From chrismo at clabs.org  Thu Oct  9 10:17:17 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 09 Oct 2003 09:17:17 -0500
Subject: [Wtr-general] Welcome
Message-ID: <3F856DED.2030302@clabs.org>

We all here?

-- 

Chris
http://clabs.org/blogki




From jkohl at telusplanet.net  Thu Oct  9 10:49:08 2003
From: jkohl at telusplanet.net (jkohl at telusplanet.net)
Date: Thu,  9 Oct 2003 07:49:08 -0700
Subject: [Wtr-general] Welcome
In-Reply-To: <3F856DED.2030302@clabs.org>
References: <3F856DED.2030302@clabs.org>
Message-ID: <1065710948.3f857564c2f30@webmail.telusplanet.net>

> We all here?
I'm in.



From paul.rogers at shaw.ca  Thu Oct  9 10:54:11 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Thu, 09 Oct 2003 08:54:11 -0600
Subject: [Wtr-general] Welcome
Message-ID: <f396e8f3be1f.f3be1ff396e8@shaw.ca>

me too.

----- Original Message -----
From: jkohl at telusplanet.net
Date: Thursday, October 9, 2003 8:49 am
Subject: Re: [Wtr-general] Welcome

> > We all here?
> I'm in.
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From chrismo at clabs.org  Thu Oct  9 11:32:37 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 09 Oct 2003 10:32:37 -0500
Subject: [Wtr-general] Syntax
Message-ID: <3F857F95.1020308@clabs.org>

The latest posts on the prior thread were focusing on the look of the 
syntax. I don't have time to further the thread right now, but I thought 
I'd get us all caught up on the list.

Paul said:

>I think a lot of people (especially in QA group as opposed to a development group) would feel daunted by the google example that Chris used - whereas a developer would probably prefer the way CHris does it.
>
>So Im wondering if we could build the library Im using on top of Chriss code to avoid redisigning the wheel, but having 2 libraries aimed at slightly different users.
>

Bret said:

> My prejudice is towards the iec.form.object.action() syntax. This is 
> more like SilkTest. I would describe Paul's as being more in the 
> Visual Test or Win Runner style.
>
> I really like the tests. For some reason, i was thinking that they 
> would have to be harder to write. I don't understand why errors are 
> being passed as an element of a doubled return value. Wouldn't it be 
> cleaner to raise them as exceptions? If i liked having to remember to 
> check return codes, i could stick with WinRunner. 


Jonathan said:

>I like that syntax as well. I've spent most of my time in the SilkTest world
>as well, so that could explain the appeal.
>
>I also like the idea of using Test:Unit as a harness. I've had to hack
>together my own in SilkTest and it wasn't easy to get things to work the way
>I'd like.
>

-- 

Chris
http://clabs.org/blogki




From jkohl at telusplanet.net  Thu Oct  9 12:24:36 2003
From: jkohl at telusplanet.net (jkohl at telusplanet.net)
Date: Thu,  9 Oct 2003 09:24:36 -0700
Subject: [Wtr-general] Syntax
In-Reply-To: <3F857F95.1020308@clabs.org>
References: <3F857F95.1020308@clabs.org>
Message-ID: <1065716676.3f858bc4e2c2d@webmail.telusplanet.net>

Quoting Chris Morris <chrismo at clabs.org>:
> The latest posts on the prior thread were focusing on the look of the 
> syntax. I don't have time to further the thread right now, but I thought 
> I'd get us all caught up on the list.
> 
> Paul said:
> I think a lot of people (especially in QA group as opposed to a
> development
> group) would feel daunted by the google example that Chris used - whereas
> a developer would probably prefer the way CHris does it.

Paul raises an important point - test case code readability. When I develop test
cases, I have a trade off between elegant script code and code that is human
readable. From a tester's perspective, I will go for code that is readable so I
(or others) can tell what the test case is actually testing. 

One of the important motivators for test automation is change detection. We want
to detect changes in the product under test and spurious errors as quickly as
possible. One of the biggest weaknesses with test automation projects that I've
seen is the inability to report defects revealed by the automated test suite
quickly. This can be difficult with automated tests unless they are designed for
quick reporting turnaround up-front.

I have developed automated test suites for QA departments that didn't have
automation specialists, or they had purchased a tool and didn't have the
knowledge of how to get that tool to work. In this type of situation, I have to
develop test cases that are readable by people who may not be strong
programmers, may not know how the tool works, etc. They need to be able to use
the test suite after I'm gone, so I develop test cases with that in mind. A good
programmer would probably cringe at my script code, but the testers can read the
code and figure out what is going on.

In one situation, a QA department had six years worth of SilkTest scripts
accumulated. They brought me in because they were having problems with reporting
turnaround times. If the automated suite ran overnight and uncovered a possible
defect, it took anywhere from 3 to 5 days for QA to file a bug report. I audited
the test suite and found that the code was elegantly programmed, but there was
no clear testing focus. Test cases were very large and it would take hours to
hand trace through them to track down spurious errors. There was so much time
invested in the test suite, and management put so much pressure on the testers
to get the suite to pass before shipping the product, the focus went from using
the suite to find bugs to getting the test suite to run. Testers actually coded
around bugs to get the suite to run. Management wanted to see an ROI on the $60
000++ they had invested in the tool. This is another reason why I think a free
tool has a lot of merit.

To address this, I had the testers adopt a development strategy based on quick
turnaround and a clear focus on test cases. Test cases had to have a clear,
singular goal. Instead of a handful of huge test cases, we would develop many
smaller test cases. The design began with a procedural manual test case with a
testing purpose, the steps to execute and the expected results. 

While I don't think this is a good practice for manual test case design (testers
on the bus syndrome), this is the only way I could figure out how to have the
tests easily readable. The steps were included in the comments at the top of the
test case, and helped the developer focus on the unit under test. Each step
would be printed out to the results file, as well as the expected results. Under
the actual results section, the verification code would be executed by the test
tool. Using this technique of many small test cases with the procedural steps
included in the test case, I was able to reduce defect reporting times on my
projects to anywhere between a few minutes to a couple of hours.

I don't think this is necessary to develop automated test cases all the time,
but I find it helpful to keep me focused when developing. If I can at least make
the code readable as a test case, I can easily trace through and report a
defect. I find that with web apps, my tests break frequently, so I use Agile
techniques when developing. (I hope that the fragility of GUI testing will be
reduced with Ruby.) I immediately go into maintenance mode and have to
frequently refactor test cases over a release cycle. The readability of my code
is extremely important to keep tests relevant.





From chrismo at clabs.org  Fri Oct 10 10:04:11 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 09:04:11 -0500
Subject: [Wtr-general] MozAxc
Message-ID: <3F86BC5B.10800@clabs.org>

I played around with the Mozilla stuffs this morning ... minor 
disappointment -- this isn't exactly what I thought it was. It's not the 
whole browser, rather it's the browser control. Details here: 
http://www.clabs.org/wtr/index.cgi?page=/MozAxc

-- 

Chris
http://clabs.org/blogki




From paul.rogers at shaw.ca  Fri Oct 10 10:31:50 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Fri, 10 Oct 2003 08:31:50 -0600
Subject: [Wtr-general] MozAxc
Message-ID: <1034a35103576f.103576f1034a35@shaw.ca>

I gave it a quick try too. Similar results to what Chris identified, however the mozilla version I had did have the dlls - just in a different directory 

C:\Program Files\Common Files\mozilla.org\GRE\1.5b_2003082704

I registered the dlls, and tried adding them to a VB form in Excel. Excel and the vb editor both locked up.

I managed to view the methods the control contains using simple ole browser (http://homepage1.nifty.com/markey/ruby/win32ole/index_e.html)
and this showed that the control had a similar structure to the IE control.

Paul

----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Friday, October 10, 2003 8:04 am
Subject: [Wtr-general] MozAxc

> I played around with the Mozilla stuffs this morning ... minor 
> disappointment -- this isn't exactly what I thought it was. It's 
> not the 
> whole browser, rather it's the browser control. Details here: 
> http://www.clabs.org/wtr/index.cgi?page=/MozAxc
> 
> -- 
> 
> Chris
> http://clabs.org/blogki
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From chrismo at clabs.org  Fri Oct 10 10:58:50 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 09:58:50 -0500
Subject: [Wtr-general] MozAxc
In-Reply-To: <1034a35103576f.103576f1034a35@shaw.ca>
References: <1034a35103576f.103576f1034a35@shaw.ca>
Message-ID: <3F86C92A.9070502@clabs.org>

paul.rogers at shaw.ca wrote:

>I gave it a quick try too. Similar results to what Chris identified, however the mozilla version I had did have the dlls - just in a different directory 
>
>C:\Program Files\Common Files\mozilla.org\GRE\1.5b_2003082704
>
D'oh! I wasn't reading closely enough. There's a mozctl.dll and 
mozctlx.dll -- only the second one should be registered. Unregistered 
the .dlls that I manually put in from the zip. Went to Common\GRE\yada, 
and registered mozctl*x*.dll, this time, I had better luck -- a little 
bit anyway.

    C:\>irb -r win32ole
    irb(main):001:0> moz = WIN32OLE.new("Mozilla.Browser")
    WIN32OLERuntimeError: Unknown OLE server : `Mozilla.Browser'
        HRESULT error code:0x800401f3
          Invalid class string
            from (irb):1:in `initialize'
            from (irb):1:in `new'
            from (irb):1
    irb(main):002:0> moz = WIN32OLE.new("Mozilla.Browser.1")
    => #<WIN32OLE:0x2b9a1c8>
    irb(main):003:0> moz.visible = true
    WIN32OLERuntimeError:
        OLE error code:8000FFFF in Mozilla.Browser.1
          Browser is not in a valid state
        HRESULT error code:0x80020009
          Exception occurred
            from (irb):3:in `method_missing'
            from (irb):3
    irb(main):004:0> moz.gohome
    WIN32OLERuntimeError: gohome
        OLE error code:8000FFFF in Mozilla.Browser.1
          Browser is not in a valid state
        HRESULT error code:0x80020009
          Exception occurred
            from (irb):4:in `method_missing'
            from (irb):4

So, dunno way you have to use the version specific ProgID to get it to 
work ... but, anyway, not sure about the Browser is not in valid state, 
except that maybe this is because the control is expecting to be 
embedded in a window, not be run standalone.

-- 

Chris
http://clabs.org/blogki




From chrismo at clabs.org  Fri Oct 10 11:01:34 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 10:01:34 -0500
Subject: [Wtr-general] List replies
Message-ID: <3F86C9CE.9020300@clabs.org>

I'd setup to list to default replies to the list instead of poster, but 
it seems to be stuck on replies set to poster ... but seems y'all have 
figured this out already.

So, Reply To All for now.

-- 

Chris
http://clabs.org/blogki




From chrismo at clabs.org  Fri Oct 10 11:06:07 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 10:06:07 -0500
Subject: [Wtr-general] List replies
In-Reply-To: <3F86C9CE.9020300@clabs.org>
References: <3F86C9CE.9020300@clabs.org>
Message-ID: <3F86CADF.9090004@clabs.org>

Chris Morris wrote:

> I'd setup to list to default replies to the list instead of poster, 
> but it seems to be stuck on replies set to poster ... but seems y'all 
> have figured this out already.
>
> So, Reply To All for now. 

Well, I reset it ... now it (normal Reply) seems to be replying to both 
the poster and the list. If this bugs anyone, lemme know and I'll follow 
up with the rubyforge folk.

-- 

Chris
http://clabs.org/blogki




From chrismo at clabs.org  Fri Oct 10 11:25:28 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 10:25:28 -0500
Subject: [Wtr-general] MozAxc
In-Reply-To: <3F86C92A.9070502@clabs.org>
References: <1034a35103576f.103576f1034a35@shaw.ca>
	<3F86C92A.9070502@clabs.org>
Message-ID: <3F86CF68.40009@clabs.org>

Chris Morris wrote:

> paul.rogers at shaw.ca wrote:
>
>> I gave it a quick try too. Similar results to what Chris identified, 
>> however the mozilla version I had did have the dlls - just in a 
>> different directory
>> C:\Program Files\Common Files\mozilla.org\GRE\1.5b_2003082704
>>
> D'oh! I wasn't reading closely enough. There's a mozctl.dll and 
> mozctlx.dll -- only the second one should be registered. Unregistered 
> the .dlls that I manually put in from the zip. Went to 
> Common\GRE\yada, and registered mozctl*x*.dll, this time, I had better 
> luck -- a little bit anyway. 

Tried from Delphi, too, and it just hung Delphi this time. Hmmm.

-- 

Chris
http://clabs.org/blogki




From bret at pettichord.com  Fri Oct 10 14:57:06 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Fri, 10 Oct 2003 13:57:06 -0500
Subject: [Wtr-general] MozAxc
In-Reply-To: <3F86BC5B.10800@clabs.org>
Message-ID: <5.1.0.14.2.20031010135606.020efb90@127.0.0.1>

At 09:04 AM 10/10/2003, Chris Morris wrote:
>I played around with the Mozilla stuffs this morning ... minor 
>disappointment -- this isn't exactly what I thought it was. It's not the 
>whole browser, rather it's the browser control. Details here: 
>http://www.clabs.org/wtr/index.cgi?page=/MozAxc

Thats a good report. Thanks for the research and for keeping us posted.

(P.S. how can a com server work unless you register it? i don't understand 
why they would tell you not to register it.)


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Fri Oct 10 16:54:08 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 10 Oct 2003 15:54:08 -0500
Subject: [Wtr-general] MozAxc
In-Reply-To: <5.1.0.14.2.20031010135606.020efb90@127.0.0.1>
References: <5.1.0.14.2.20031010135606.020efb90@127.0.0.1>
Message-ID: <3F871C70.5070408@clabs.org>

Bret Pettichord wrote:

>> I played around with the Mozilla stuffs this morning ... minor 
>> disappointment -- this isn't exactly what I thought it was. It's not 
>> the whole browser, rather it's the browser control. Details here: 
>> http://www.clabs.org/wtr/index.cgi?page=/MozAxc
>
>
> Thats a good report. Thanks for the research and for keeping us posted.
>
> (P.S. how can a com server work unless you register it? i don't 
> understand why they would tell you not to register it.) 

I misread their home page, haven't updated the Wiki yet. There's two 
.dlls mozctl and mozctlx. They say to register mozctlx and warn not to 
register the other one. I didn't ever see the 'x' distinguishing between 
the two.

-- 

Chris
http://clabs.org/blogki




From jkohl at telusplanet.net  Sat Oct 11 14:51:58 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Sat, 11 Oct 2003 12:51:58 -0600
Subject: [Wtr-general] thoughts on the focus of this project
Message-ID: <000001c39028$bf935020$6401a8c0@tintin>

We seem to be having some issues deciding on exactly what we all want to do
with this project. I think restricting the focus is the best strategy.  If
people want, we could set up a skeleton overarching project for future use,
with this as the first and only actual project going on at the moment. Also,
I think we have to be careful that we aren't trying to be everything to
everyone. That will doom us to failure, because we won't do anything well
without a clear focus.

 

To begin with, we may want to focus on web testing with Ruby, and decide
what that project's goal should be. My personal goal is to have a toolkit of
Ruby functions to draw from to use for writing test cases for web
applications. Integrating that toolkit with Test::Unit would be a great fit,
as that would be the natural test harness to use as a Ruby developer.

 

>From my perspective, this would involve building more of a library for
testing rather than some kind of test environment. This library should be
expandable.  For example, if we choose to diversify into .Net forms, that
could be another part of the library. If down the road we feel something
like this would work for non-web applications, we could start another
project under the larger project umbrella.

 

My point is that we should choose to do very few things, and do them well,
and then expand later as need demands. I like to follow the XP tenet of not
trying to do everything up front.

 

Does this gel with all of you and your thoughts and goals for this project?
Did I miss the boat and get off on a tangent here?

 

Thanks;

 

-Jonathan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20031011/8c1c2c36/attachment.html 

From bret at pettichord.com  Sat Oct 11 15:20:39 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sat, 11 Oct 2003 14:20:39 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <000001c39028$bf935020$6401a8c0@tintin>
Message-ID: <5.1.0.14.2.20031011141848.03365810@127.0.0.1>

I like the idea of focus. I suggest that we actually target a specific web 
application and build a demonstration testsuite (or mulitiple testsuites, 
perhaps exemplifying different approaches). Maybe this could the timeclock 
application that Brian developed. Or maybe we could have a couple.

Bret

At 01:51 PM 10/11/2003, Jonathan Kohl wrote:

>We seem to be having some issues deciding on exactly what we all want to 
>do with this project. I think restricting the focus is the best 
>strategy.  If people want, we could set up a skeleton overarching project 
>for future use, with this as the first and only actual project going on at 
>the moment. Also, I think we have to be careful that we aren't trying to 
>be everything to everyone. That will doom us to failure, because we won't 
>do anything well without a clear focus.
>
>
>
>To begin with, we may want to focus on web testing with Ruby, and decide 
>what that project s goal should be. My personal goal is to have a toolkit 
>of Ruby functions to draw from to use for writing test cases for web 
>applications. Integrating that toolkit with Test::Unit would be a great 
>fit, as that would be the natural test harness to use as a Ruby developer.
>
>
>
> From my perspective, this would involve building more of a library for 
> testing rather than some kind of test environment. This library should be 
> expandable.  For example, if we choose to diversify into .Net forms, that 
> could be another part of the library. If down the road we feel something 
> like this would work for non-web applications, we could start another 
> project under the larger project umbrella.
>
>
>
>My point is that we should choose to do very few things, and do them well, 
>and then expand later as need demands. I like to follow the XP tenet of 
>not trying to do everything up front.
>
>
>
>Does this gel with all of you and your thoughts and goals for this 
>project? Did I miss the boat and get off on a tangent here?
>
>
>
>Thanks;
>
>
>
>-Jonathan
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Sun Oct 12 18:53:40 2003
From: chrismo at clabs.org (Chris Morris)
Date: Sun, 12 Oct 2003 17:53:40 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <000001c39028$bf935020$6401a8c0@tintin>
References: <000001c39028$bf935020$6401a8c0@tintin>
Message-ID: <3F89DB74.2050502@clabs.org>

Jonathan Kohl wrote:

> Does this gel with all of you and your thoughts and goals for this 
> project? Did I miss the boat and get off on a tangent here?
>
No, that works for me.

I'm the one who got off on broadening the scope, mainly because I wanted 
a good name and could easily see how "Web" and "Ruby" might be too 
restrictive. But I knew I needed y'alls feedback, and I've gotten that 
-- so certainly sticking to WTR is fine with me, and starting small is 
always a good idea, especially with volunteer projects like this, which 
will probably go in spurts.

-- 
Chris
http://clabs.org/blogki





From jkohl at telusplanet.net  Sun Oct 12 19:01:57 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Sun, 12 Oct 2003 17:01:57 -0600
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <3F89DB74.2050502@clabs.org>
Message-ID: <000001c39114$d620c0b0$6401a8c0@tintin>

> No, that works for me.
Good that we're on the same page.

> I'm the one who got off on broadening the scope, mainly because I wanted
> a good name and could easily see how "Web" and "Ruby" might be too
> restrictive.
Nah -we all have ideas on where this should go. A common thread will
hopefully emerge as things progress. I see a couple things that might
happen.
1. Due to differing backgrounds and applications of a tool like this, we
could end up with a "everyone design independently, compare 
designs, merge/select best approach/ideas" situation which could work.
2. We might be completely off the first time implementing WTR against an
example. The end result might be a moving target, but if we all move in that
direction together, we should be able to hit it.
To quote Fred Brooks and ESR, ""Plan to throw one away; you will, anyhow."
(Fred Brooks, the Mythical Man-Month, chapter 11)
Or, to put it another way, you often don't really understand the problem
until after the first time you implement a solution. The second time, maybe
you know enough to do it right. So if you want to get it right, be ready to
start over at least once."






From jkohl at telusplanet.net  Sun Oct 12 19:02:11 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Sun, 12 Oct 2003 17:02:11 -0600
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <5.1.0.14.2.20031011141848.03365810@127.0.0.1>
Message-ID: <000101c39114$de6e4d50$6401a8c0@tintin>

> I like the idea of focus. I suggest that we actually target a specific web
> application and build a demonstration testsuite (or mulitiple testsuites,
> perhaps exemplifying different approaches). Maybe this could the timeclock
> application that Brian developed. Or maybe we could have a couple.
I like the idea of targeting an application to design around. If we write
the WTR code independently of an actual testing situation, we might miss the
target.





From bret at pettichord.com  Sun Oct 12 14:48:55 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sun, 12 Oct 2003 13:48:55 -0500
Subject: [Wtr-general] closing and then creating iec
Message-ID: <5.1.0.14.2.20031012134802.03598630@127.0.0.1>

The following fails intermittently
     ...
     @iec.close
     @iec = ClIEController.new(VISIBLE)

Adding a "sleep(1)" between the calls corrects. Is there a better way?

(Complete demonstration tests are attached.)


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com
-------------- next part --------------
# Demonstrate a bug in restarting the IE Controller.

require 'cl/iec'
require 'test/unit'

VISIBLE = TRUE

class RestartTests < Test::Unit::TestCase
  def set_up ()
    "Start on the login page."
    @iec = ClIEController.new(VISIBLE)
    @iec.navigate ("http://localhost:8080")
  end
    
  def tear_down ()
    "Shutdown browser."
    @iec.close
    #sleep(.1) # uncomment to avoid bug
  end

  def test_null()
  end

  def test_null2()
  end

  def test_null3()
  end

end

# Sometimes the following error will occur with the second test. For me it happens more often than not.

#Error occurred in test_null2(TimeClockLoginTests): WIN32OLERuntimeError: Unknown
# property or method : `busy'
#    HRESULT error code:0x800706be
#      The remote procedure call failed
#        c:/ruby/lib/ruby/site_ruby/1.6/cl/iec/cliecontroller.rb:165:in `method_m
#issing'
#        c:/ruby/lib/ruby/site_ruby/1.6/cl/iec/cliecontroller.rb:165:in `waitForI
#E'
#        c:/ruby/lib/ruby/site_ruby/1.6/cl/iec/cliecontroller.rb:136:in `navigate
#'
#        ie-timeclock-tests.rb:10:in `set_up'
#        ie-timeclock-tests.rb:49















From bret at pettichord.com  Mon Oct 13 01:21:18 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Mon, 13 Oct 2003 00:21:18 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <000101c39114$de6e4d50$6401a8c0@tintin>
References: <5.1.0.14.2.20031011141848.03365810@127.0.0.1>
Message-ID: <5.1.0.14.2.20031013001857.02775268@127.0.0.1>

At 06:02 PM 10/12/2003, Jonathan Kohl wrote:
>I like the idea of targeting an application to design around. If we write
>the WTR code independently of an actual testing situation, we might miss the
>target.

Well, i think right now we are all writing against different targets -- 
different apps. It would be good to have a common app we could write tests 
against.

I agree that we will have different ideas about what kinds of idioms the 
resulting tests will be written in. I'm happy to see and compare different 
ideas.

I would like to have a common goal in terms of they kind of application we 
are testing and the basic approach (i.e. driving IE). I'd like to see a 
variety of design ideas.

Bret



_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Mon Oct 13 01:18:33 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Mon, 13 Oct 2003 00:18:33 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <000001c39114$d620c0b0$6401a8c0@tintin>
References: <3F89DB74.2050502@clabs.org>
Message-ID: <5.1.0.14.2.20031013001803.0273e7f8@127.0.0.1>

At 06:01 PM 10/12/2003, Jonathan Kohl wrote:
>"Plan to throw one away; you will, anyhow."

Didn't Paul just do that with the Digital Oil Works code base?


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Mon Oct 13 08:42:11 2003
From: chrismo at clabs.org (Chris Morris)
Date: Mon, 13 Oct 2003 07:42:11 -0500
Subject: [Wtr-general] closing and then creating iec
In-Reply-To: <5.1.0.14.2.20031012134802.03598630@127.0.0.1>
References: <5.1.0.14.2.20031012134802.03598630@127.0.0.1>
Message-ID: <3F8A9DA3.1080900@clabs.org>

Bret Pettichord wrote:

> The following fails intermittently
>     ...
>     @iec.close
>     @iec = ClIEController.new(VISIBLE)
>
> Adding a "sleep(1)" between the calls corrects. Is there a better way?

Huh - that's strange. Only better way that comes to mind is to somehow 
stick this sleeping bit into cliec's close method. I wonder if calling 
cliec's wait method would also do the trick (I'm guessing it'd fail, 
though). Maybe a call to wait inside cliec's initialize method. 
Obviously, something we'd have to experiment with.



From chrismo at clabs.org  Mon Oct 13 08:44:46 2003
From: chrismo at clabs.org (Chris Morris)
Date: Mon, 13 Oct 2003 07:44:46 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <5.1.0.14.2.20031013001857.02775268@127.0.0.1>
References: <5.1.0.14.2.20031011141848.03365810@127.0.0.1>
	<5.1.0.14.2.20031013001857.02775268@127.0.0.1>
Message-ID: <3F8A9E3E.9020303@clabs.org>

Bret Pettichord wrote:

> I would like to have a common goal in terms of they kind of 
> application we are testing and the basic approach (i.e. driving IE). 
> I'd like to see a variety of design ideas.

We've also mentioned to need for cliec to have tests for itself -- 
obviously this would entail some sort of example application to drive 
cliec over.



From jkohl at telusplanet.net  Mon Oct 13 12:02:11 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Mon, 13 Oct 2003 10:02:11 -0600
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <5.1.0.14.2.20031013001803.0273e7f8@127.0.0.1>
Message-ID: <000901c391a3$5c867bc0$6401a8c0@tintin>


> >"Plan to throw one away; you will, anyhow."
> Didn't Paul just do that with the Digital Oil Works code base?
Yes - I didn't think of that.




From paul.rogers at shaw.ca  Mon Oct 13 22:24:30 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Mon, 13 Oct 2003 20:24:30 -0600
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <000901c391a3$5c867bc0$6401a8c0@tintin>
Message-ID: <000601c391fa$4ca3a370$7aa09044@NewDell>

Its almost thrown away.

It gave me ideas for a new one though. Implementing it as an object
rather than functions as in the google example will allow features to be
added easier. I started doing this in the example I sent round a week of
so ago.

Paul

-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Jonathan Kohl
Sent: 13 October 2003 09:02
To: wtr-general at rubyforge.org
Subject: RE: [Wtr-general] thoughts on the focus of this project



> >"Plan to throw one away; you will, anyhow."
> Didn't Paul just do that with the Digital Oil Works code base?
Yes - I didn't think of that.


_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From bret at pettichord.com  Tue Oct 14 00:50:21 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Mon, 13 Oct 2003 23:50:21 -0500
Subject: [Wtr-general] thoughts on the focus of this project
In-Reply-To: <3F8A9E3E.9020303@clabs.org>
References: <5.1.0.14.2.20031013001857.02775268@127.0.0.1>
	<5.1.0.14.2.20031011141848.03365810@127.0.0.1>
	<5.1.0.14.2.20031013001857.02775268@127.0.0.1>
Message-ID: <5.1.0.14.2.20031013234917.02649d60@127.0.0.1>

At 07:44 AM 10/13/2003, Chris Morris wrote:
>We've also mentioned to need for cliec to have tests for itself -- 
>obviously this would entail some sort of example application to drive 
>cliec over.

I now realize that we can just load html files to drive many types of 
tests. That simplifies this immensely.



_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Tue Oct 14 00:48:57 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Mon, 13 Oct 2003 23:48:57 -0500
Subject: [Wtr-general] closing and then creating iec
In-Reply-To: <3F8A9DA3.1080900@clabs.org>
References: <5.1.0.14.2.20031012134802.03598630@127.0.0.1>
	<5.1.0.14.2.20031012134802.03598630@127.0.0.1>
Message-ID: <5.1.0.14.2.20031013234818.0264aee8@127.0.0.1>

i did try cliec.wait, but as you expected, it failed.

At 07:42 AM 10/13/2003, Chris Morris wrote:
>Bret Pettichord wrote:
>
>>The following fails intermittently
>>     ...
>>     @iec.close
>>     @iec = ClIEController.new(VISIBLE)
>>
>>Adding a "sleep(1)" between the calls corrects. Is there a better way?
>
>Huh - that's strange. Only better way that comes to mind is to somehow 
>stick this sleeping bit into cliec's close method. I wonder if calling 
>cliec's wait method would also do the trick (I'm guessing it'd fail, 
>though). Maybe a call to wait inside cliec's initialize method. Obviously, 
>something we'd have to experiment with.
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Tue Oct 14 11:07:23 2003
From: chrismo at clabs.org (Chris Morris)
Date: Tue, 14 Oct 2003 10:07:23 -0500
Subject: [Wtr-general] cliec in rubyforge
Message-ID: <3F8C112B.7070902@clabs.org>

Woop. I got the stuff moved over. http://rubyforge.org/scm/?group_id=104

I did a quick re-org of the source directories ... so the build.rb 
script probably is borken now, but I'll be getting it cleaned up in 
hopefully the next few days and try to get a release out the door: (1) 
for the dependency removal (2) just to get it done in the new home and 
work out any release kinks.

-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Tue Oct 14 13:11:15 2003
From: chrismo at clabs.org (Chris Morris)
Date: Tue, 14 Oct 2003 12:11:15 -0500
Subject: [Wtr-general] cliec in rubyforge
In-Reply-To: <3F8C112B.7070902@clabs.org>
References: <3F8C112B.7070902@clabs.org>
Message-ID: <3F8C2E33.1050002@clabs.org>

Chris Morris wrote:

> Woop. I got the stuff moved over. http://rubyforge.org/scm/?group_id=104

BTW, checkout http://sfsetup.sf.net for a Windows program that will 
setup the right SSH client (and WinCVS, apparently) for working with 
RubyForge.net.

PuTTY does not (or I couldn't get it to) work with RubyForge.net.

-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Tue Oct 14 17:06:37 2003
From: chrismo at clabs.org (Chris Morris)
Date: Tue, 14 Oct 2003 16:06:37 -0500
Subject: [Wtr-general] [XP] RFC: Acceptance Test Tools
Message-ID: <3F8C655D.9070403@clabs.org>

Hooked this guy off the XP list this afternoon ... gave him my 2 cents 
and told him to get on the mailing list to get y'alls feedback.

-------- Original Message --------
Subject: 	RE: [XP] RFC: Acceptance Test Tools
Date: 	Tue, 14 Oct 2003 16:15:29 -0400
From: 	Thomas Robbs <Thomas.Robbs at casero.com>
To: 	<chrismo at clabs.org>



Hi Chris:

Thanks for the reply.  We're supporting Mozilla and IE.

What benefits do you find you get out of driving IE with Ruby over the
other tools I mentioned?  I.e. what drove you to create this over using
one of the existing tools that is available?

Thanks again for the link, and stay in touch.

Cheers,

Tom

> -----Original Message-----
> From: Chris Morris [mailto:chrismo at clabs.org] 
> Sent: October 14, 2003 4:07 PM
> To: Thomas Robbs
> Subject: RE: [XP] RFC: Acceptance Test Tools
> 
> 
> >
> >
> >I'm currently looking for an Acceptance Test tool which will 
> help our 
> >team build automated acceptance tests.  A user accesses and uses our 
> >product by way of a web browser, however the product may be 
> deployed in 
> >an application server in the operator's NOC or on a CPE 
> device in the 
> >home.  Either way, the user still uses a web browser to use the 
> >services provided by our software.  This is the first area 
> which I'll 
> >focus acceptance testing on.
> >
> I'm replying directly to you since I'm not really answering your 
> question ...
> 
> I assume you'll be supporting more browsers than IE, but you 
> may want to 
> this out[1]. Basically it's a Ruby lib to drive Internet 
> Explorer, for 
> any purpose really, but primarily for acceptance testing.
> 
> We're trying to get Mozilla support -- it might be close at 
> hand, but we 
> aren't sure yet. In addition, if you have need for a mature 
> tool, this 
> ain't it. However, if you're in a position to use it and 
> contribute to 
> it, you'd be more than welcome.
> 
> [1] http://clabs.org/wtr
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> 



-- 
Chris
http://clabs.org/blogki




From paul.rogers at shaw.ca  Tue Oct 14 17:45:01 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Tue, 14 Oct 2003 15:45:01 -0600
Subject: [Wtr-general] Ruby and IE
Message-ID: <12bf25112bf254.12bf25412bf251@shaw.ca>

Tom,

hope you dont mind me sending to you directly, but wasnt sure if you were on the list.

Both myself and Chris have developed different testing tools for what you describe. Chriss one is available on the rubyforge site - I'll send you a sample of mine later - Its still in development right now - as the original is owned by my employer.

We took slightly different appraches - I got quite a lot of inspiraion fromChris, and even stole some code! 

I have been amazed at what I can do with ruby and IE - the level of automation we have acheived, and the time in which we have done it is incredible. It was probably 1/2 the time it would have taken using silk or visual test ( both of which Ive used in previous jobs)


Paul




From chrismo at clabs.org  Wed Oct 15 08:14:07 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 07:14:07 -0500
Subject: [Wtr-general] Paul's code
In-Reply-To: <12bf25112bf254.12bf25412bf251@shaw.ca>
References: <12bf25112bf254.12bf25412bf251@shaw.ca>
Message-ID: <3F8D3A0F.5090102@clabs.org>

paul.rogers at shaw.ca wrote:

>Both myself and Chris have developed different testing tools for what you describe. Chriss one is available on the rubyforge site - I'll send you a sample of mine later - Its still in development right now - as the original is owned by my employer.
>  
>
I didn't realize you were working on rewriting your stuff. Can you share 
more about what you're doing?

>We took slightly different appraches - I got quite a lot of inspiraion fromChris, and even stole some code! 
>  
>
... it wasn't stealing, it was open-sourced :-)

-- 
Chris
http://clabs.org/blogki




From paul.rogers at shaw.ca  Wed Oct 15 09:37:45 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Wed, 15 Oct 2003 07:37:45 -0600
Subject: [Wtr-general] Paul's code
Message-ID: <5e90d624f3.624f35e90d@shaw.ca>

I sent some out a week or so ago. It was a button method and a radio button method. I'll send it on again tonight if you dont have it.

----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Wednesday, October 15, 2003 6:14 am
Subject: [Wtr-general] Paul's code

> paul.rogers at shaw.ca wrote:
> 
> >Both myself and Chris have developed different testing tools for 
> what you describe. Chriss one is available on the rubyforge site - 
> I'll send you a sample of mine later - Its still in development 
> right now - as the original is owned by my employer.
> >  
> >
> I didn't realize you were working on rewriting your stuff. Can you 
> share 
> more about what you're doing?
> 
> >We took slightly different appraches - I got quite a lot of 
> inspiraion fromChris, and even stole some code! 
> >  
> >
> ... it wasn't stealing, it was open-sourced :-)
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From chrismo at clabs.org  Wed Oct 15 11:35:21 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 10:35:21 -0500
Subject: [Wtr-general] Paul's code
In-Reply-To: <5e90d624f3.624f35e90d@shaw.ca>
References: <5e90d624f3.624f35e90d@shaw.ca>
Message-ID: <3F8D6939.9020402@clabs.org>

paul.rogers at shaw.ca wrote:

>I sent some out a week or so ago. It was a button method and a radio button method. I'll send it on again tonight if you dont have it.
>  
>
I saw the snippet in the document attachment you emailed out before. 
Sounds like the same thing.

 From what I saw, it looks like you're going in a different direction 
syntactically than what I've already got -- which in and of itself is 
fine, we still need to continue the syntax discussion, I think. My 
question is, how much effort are you putting into rewriting what you had 
at Digital OilField, and is that going to leave us with two pieces of 
code that we then later want to merge? If so, then I think it'd be time 
better spent to work towards a merged contribution from the get go.

-- 
Chris
http://clabs.org/blogki




From kingsley at icecode.org  Wed Oct 15 14:35:12 2003
From: kingsley at icecode.org (Kingsley)
Date: Wed, 15 Oct 2003 19:35:12 +0100
Subject: [Wtr-general] Paul's code
In-Reply-To: <5e90d624f3.624f35e90d@shaw.ca>
References: <5e90d624f3.624f35e90d@shaw.ca>
Message-ID: <200310151935.12598.kingsley@icecode.org>

Hi 

I'm keen to help out on your Ruby test tool.

I'm learning Ruby - I'm not a developer yet - I'm a tester by profession.

My Ruby is getting better and if you guys are coding in TDD then I'll 
definitely be up for helping.

How do I find out what needs doing ?

Thanks

Kingsley



From paul.rogers at shaw.ca  Wed Oct 15 14:59:01 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Wed, 15 Oct 2003 12:59:01 -0600
Subject: [Wtr-general] Paul's code
Message-ID: <c5e03c597b.c597bc5e03@shaw.ca>

The reason Ive gone in the different direction is the maintenance issue - I think the cLabs code is more suited to developer unit testing whereas mine is more suited to 'qa group' testing. 

This is because mine more closely resembles what a manual tester (clickbutton, clickLink etc) would do, which would make writing a script for a non-developer that much easier.

Ideally I would like to add this as a layer on top of the cLabs code so we have just the single code base, although Im not too sure how this would work out right now.

I'll take a look at Test::Unit and see how my code will work with that.

Paul



----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Wednesday, October 15, 2003 9:35 am
Subject: Re: [Wtr-general] Paul's code

> paul.rogers at shaw.ca wrote:
> 
> >I sent some out a week or so ago. It was a button method and a 
> radio button method. I'll send it on again tonight if you dont 
> have it.
> >  
> >
> I saw the snippet in the document attachment you emailed out 
> before. 
> Sounds like the same thing.
> 
> From what I saw, it looks like you're going in a different 
> direction 
> syntactically than what I've already got -- which in and of itself 
> is 
> fine, we still need to continue the syntax discussion, I think. My 
> question is, how much effort are you putting into rewriting what 
> you had 
> at Digital OilField, and is that going to leave us with two pieces 
> of 
> code that we then later want to merge? If so, then I think it'd be 
> time 
> better spent to work towards a merged contribution from the get go.
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From chrismo at clabs.org  Wed Oct 15 15:03:48 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 14:03:48 -0500
Subject: [Wtr-general] Paul's code
In-Reply-To: <200310151935.12598.kingsley@icecode.org>
References: <5e90d624f3.624f35e90d@shaw.ca>
	<200310151935.12598.kingsley@icecode.org>
Message-ID: <3F8D9A14.90306@clabs.org>

Kingsley wrote:

>Hi 
>
>I'm keen to help out on your Ruby test tool.
>  
>
Excellent!

>I'm learning Ruby - I'm not a developer yet - I'm a tester by profession.
>
>My Ruby is getting better and if you guys are coding in TDD then I'll 
>definitely be up for helping.
>
>How do I find out what needs doing ?
>  
>
You can start here: http://www.clabs.org/wtr/index.cgi?page=/ClIEController

... and try to get ahold of the code that's currently available. Sorry, 
it's early enough we don't have a clear DOWNLOAD HERE link.

I've also started a ToDo page off the link I gave you -- nothing in it, 
though we need to start collecting stories in there.

Getting a good test web page together to be a mock to run the controller 
against would be among the top things.

-- 
Chris
http://clabs.org/blogki




From paul.rogers at shaw.ca  Wed Oct 15 15:05:42 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Wed, 15 Oct 2003 13:05:42 -0600
Subject: [Wtr-general] Paul's code
Message-ID: <cb09fc6ff5.c6ff5cb09f@shaw.ca>

Chris - do you still have the zip I sent last week? There is an html page in that that may serve as the basis for a test page.

----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Wednesday, October 15, 2003 1:03 pm
Subject: Re: [Wtr-general] Paul's code

> Kingsley wrote:
> 
> >Hi 
> >
> >I'm keen to help out on your Ruby test tool.
> >  
> >
> Excellent!
> 
> >I'm learning Ruby - I'm not a developer yet - I'm a tester by 
> profession.>
> >My Ruby is getting better and if you guys are coding in TDD then 
> I'll 
> >definitely be up for helping.
> >
> >How do I find out what needs doing ?
> >  
> >
> You can start here: 
> http://www.clabs.org/wtr/index.cgi?page=/ClIEController
> ... and try to get ahold of the code that's currently available. 
> Sorry, 
> it's early enough we don't have a clear DOWNLOAD HERE link.
> 
> I've also started a ToDo page off the link I gave you -- nothing 
> in it, 
> though we need to start collecting stories in there.
> 
> Getting a good test web page together to be a mock to run the 
> controller 
> against would be among the top things.
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From chrismo at clabs.org  Wed Oct 15 15:26:40 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 14:26:40 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F857F95.1020308@clabs.org>
References: <3F857F95.1020308@clabs.org>
Message-ID: <3F8D9F70.2010802@clabs.org>

I fear an impasse here. Syntax style is not going to be proven 
definitively easier one way or the other. Obviously, whatever one's own 
prior experience is will greatly influence that person's definition of 
easier. Right now there's a thread on the Ruby list about the 
readability of parentheses around parameters to a method call. It's 
getting equal hits from across the spectrum.

My concern with having two different styles is that it will make the 
library itself more unpredictable to the user. Here we do this, there we 
do that -- unless we try to offer two layers as Paul suggested. Then my 
concern is with coding time -- I'm not going to have much time to 
contribute, everything's going to be volunteer, and trying to maintain 
two styles at the same time sounds like a bit much. 

Thoughts?

-- 
Chris
http://clabs.org/blogki




From jkohl at telusplanet.net  Wed Oct 15 19:14:06 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Wed, 15 Oct 2003 17:14:06 -0600
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8D9F70.2010802@clabs.org>
Message-ID: <000001c39372$085aa960$6401a8c0@tintin>

I'm not sure if we're moving towards an impasse or not. What is the sticking
issue? At the risk of over-simplifying, is it one of these scenarios: 

1) Do we design a library around elegantly-programmed code, or around code
that testers can easily read and use?
2) Two code bases: Chris' Clabs IE Controller and Paul's IE Controller. Do
we need to choose one code base, drop the other and move forward with one?

I don't know if I want to see a forking of efforts before we really start,
so what is everyone's goal with the project? If the goals aren't congruent
with each other, then I too fear an impasse.

My goal is to have a library of functions that I can use for testing web
applications. Ideally, they should be developed so that test cases can be
written easily using them, and they should work with a harness like
Test::Unit. My bias is towards test cases that are readable by testers who
are not necessarily programmers. What that syntax will look like in the end
I'm not sure yet, but I do like the SilkTest way of doing things.

That's my view. What is everyone else's goal?

-Jonathan


> -----Original Message-----
> From: wtr-general-bounces at rubyforge.org [mailto:wtr-general-
> bounces at rubyforge.org] On Behalf Of Chris Morris
> Sent: October 15, 2003 1:27 PM
> To: chrismo at clabs.org; wtr-general at rubyforge.org
> Subject: Re: [Wtr-general] Syntax
> 
> I fear an impasse here. Syntax style is not going to be proven
> definitively easier one way or the other. Obviously, whatever one's own
> prior experience is will greatly influence that person's definition of
> easier. Right now there's a thread on the Ruby list about the
> readability of parentheses around parameters to a method call. It's
> getting equal hits from across the spectrum.
> 
> My concern with having two different styles is that it will make the
> library itself more unpredictable to the user. Here we do this, there we
> do that -- unless we try to offer two layers as Paul suggested. Then my
> concern is with coding time -- I'm not going to have much time to
> contribute, everything's going to be volunteer, and trying to maintain
> two styles at the same time sounds like a bit much.
> 
> Thoughts?
> 
> --
> Chris
> http://clabs.org/blogki
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general




From chrismo at clabs.org  Wed Oct 15 19:31:55 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 18:31:55 -0500
Subject: [Wtr-general] Paul's Attachment
Message-ID: <3F8DD8EB.4080702@clabs.org>

Here 'tis

-------- Original Message --------
Subject: 	RE: IE Controller
Date: 	Wed, 08 Oct 2003 20:41:10 -0600
From: 	Paul Rogers <paul.rogers at shaw.ca>
To: 	jkohl at telusplanet.net, chrismo at clabs.org
CC: 	'Bret Pettichord' <bret at pettichord.com>



So this is the current project Im working on. It's a very small subset
of what I have been doing at digital
oilfield.

I think it's a good starting point. The code from the paper was really
just an example of what could be done

Unzip to some place, edit the testiecontroller.rb file to navigate to
the .html doc

We'd need similar methods for check boxes, links, fields, select boxes.

Chris, Id appreciate your thoughts on the code as it stands.

Paul

-----Original Message-----
From: jkohl at telusplanet.net [mailto:jkohl at telusplanet.net] 
Sent: 08 October 2003 10:30
To: chrismo at clabs.org
Cc: paul.rogers at shaw.ca; Bret Pettichord
Subject: Re: IE Controller



> I'm thinking of starting a wtr SourceForge project where we can start
> storing source -- originally was thinking of starting a cliec project,

> but having a broader wtr name can include the cliec project and any 
> others that come along. Thoughts? (We could also setup a mailing list 
> there and get off this manual thread...)
I like the WTR moniker for this specific project, but setting up a
larger SourceForge project under a different name makes sense if it
would include other web testing projects. Open Source Web Testing OSWT
or something like that would be  great, with the WTR and other useful
subprojects under that umbrella of tools.




-- 
Chris
http://clabs.org/blogki
http://cministries.org/blogki
-=-=-=-=--=-=-=-=-=-=-
Free solo piano album (mp3)
http://cministries.org/cstudios/blackandwhite.htm

-------------- next part --------------
A non-text attachment was scrubbed...
Name: cm1.zip
Type: application/x-zip-compressed
Size: 1855 bytes
Desc: not available
Url : http://rubyforge.org/pipermail/wtr-general/attachments/20031015/6be50c3e/attachment.bin 

From chrismo at clabs.org  Wed Oct 15 19:55:12 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 18:55:12 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <000001c39372$085aa960$6401a8c0@tintin>
References: <000001c39372$085aa960$6401a8c0@tintin>
Message-ID: <3F8DDE60.70404@clabs.org>

Jonathan Kohl wrote:

>I'm not sure if we're moving towards an impasse or not. 
>
I just meant on the syntax issue -- sorry if that came across too 
strong. I wanted to further the thread and started to write a return 
reply to Paul on the syntax, then realized everyone had shared their 
opinion and this sort of issue isn't going to be resolved by someone 
else convincing the other that their opinion is more right. So I thought 
I'd try to move the thread to other issues.

Paul's thinking we can do both, maybe we can, but I have my concerns and 
raised those. If we can do both, then there's no need to continue the 
debate.

>1) Do we design a library around elegantly-programmed code, or around code
>that testers can easily read and use?
>
Well, I don't really think we should put it like that even. Surely we 
should shoot for code that's elegant and easy to read and use.

Basically, it's just been the difference between this style:

    clickButton('submit')

and this style:

    form.button.click

>2) Two code bases: Chris' Clabs IE Controller and Paul's IE Controller. Do
>we need to choose one code base, drop the other and move forward with one?
>
To me this is a separate issue. I've brought this up on the other thread 
(Paul's code) -- but this I don't think is anything that needs to be 
decided as much beforehand -- so long as we don't duplicate ourselves. I 
think it's easier to merge code together over time, given things like 
syntax style not conflicting too much.

>My goal is to have a library of functions that I can use for testing web
>applications. Ideally, they should be developed so that test cases can be
>written easily using them, and they should work with a harness like
>Test::Unit. 
>
btw, I do agree with the Test::Unit usage. That framework buys us a lot 
of small stuff we won't have to hassle with, and it's standard in the 
Ruby community.

>My bias is towards test cases that are readable by testers who
>are not necessarily programmers. What that syntax will look like in the end
>I'm not sure yet, but I do like the SilkTest way of doing things.
>
This is the tricky bit, defining what's readable and easy to use to a 
large group of people.

Perhaps I can take the Google snippet from before and work on keeping 
the syntax style I prefer and improving the readability for the 
non-developer.

-- 
Chris
http://clabs.org/blogki





From jkohl at telusplanet.net  Wed Oct 15 20:06:47 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Wed, 15 Oct 2003 18:06:47 -0600
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8DDE60.70404@clabs.org>
Message-ID: <000001c39379$64c23180$6401a8c0@tintin>


> >I'm not sure if we're moving towards an impasse or not.
> I just meant on the syntax issue -- sorry if that came across too
> strong. I wanted to further the thread and started to write a return
> reply to Paul on the syntax, then realized everyone had shared their
> opinion and this sort of issue isn't going to be resolved by someone
> else convincing the other that their opinion is more right. So I thought
> I'd try to move the thread to other issues.
So we're looking at a syntax issue - I read it as a project issue. My
mistake.

> Paul's thinking we can do both, maybe we can, but I have my concerns and
> raised those. If we can do both, then there's no need to continue the
> debate.
I see where you are coming from, and I'm not sure whether two code bases is
advantageous to the project at this stage or not.

> >1) Do we design a library around elegantly-programmed code, or around
> code
> >that testers can easily read and use?
> Well, I don't really think we should put it like that even. Surely we
> should shoot for code that's elegant and easy to read and use.
Absolutely. I just wanted to figure out what the issue was - and obviously I
was off the mark if it is more a syntax issue than a design issue. The
subject line of the email should have been my first clue. 

> Basically, it's just been the difference between this style:
> 
>     clickButton('submit')
> 
> and this style:
> 
>     form.button.click
I see. The Winrunner or Visual Test sort of approach vs. the SilkTest
approach (to oversimplify yet again). I do like the iec.form.object.action()
syntax myself, but I have spent most of my time developing with 4Test so
that is familiar to me.

So do we need to choose a style, agree and move forward at this point? Any
other thoughts with regard to syntax styles? How do we address these
concerns and move forward? 

> >2) Two code bases: Chris' Clabs IE Controller and Paul's IE Controller.
> Do
> >we need to choose one code base, drop the other and move forward with
> one?
> To me this is a separate issue. I've brought this up on the other thread
> (Paul's code) -- but this I don't think is anything that needs to be
> decided as much beforehand -- so long as we don't duplicate ourselves. I
> think it's easier to merge code together over time, given things like
> syntax style not conflicting too much.
I agree.

> btw, I do agree with the Test::Unit usage. That framework buys us a lot
> of small stuff we won't have to hassle with, and it's standard in the
> Ruby community.
Good.  :-)

> This is the tricky bit, defining what's readable and easy to use to a
> large group of people.
It is tricky, but if we all have a common goal (vague as it might be), we
can at least start somewhere and the details will hopefully emerge as the
design is tried out.

> Perhaps I can take the Google snippet from before and work on keeping
> the syntax style I prefer and improving the readability for the
> non-developer.
That sounds good to me.





From chrismo at clabs.org  Wed Oct 15 22:25:08 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 21:25:08 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <000001c39379$64c23180$6401a8c0@tintin>
References: <000001c39379$64c23180$6401a8c0@tintin>
Message-ID: <3F8E0184.6020302@clabs.org>


>I see where you are coming from, and I'm not sure whether two code bases is
>advantageous to the project at this stage or not.
>

As I've read Paul so far, he's not talking two code bases. He said he
was thinking of putting his style syntax (tester-friendly) on top of the
programmer friendly code, so that his is just a wrapper on it. That's
how I've taken it.

>>Perhaps I can take the Google snippet from before and work on keeping
>>the syntax style I prefer and improving the readability for the
>>non-developer.
>>    
>>
>That sounds good to me.
>  
>

-- 
Chris
http://clabs.org/blogki





From chrismo at clabs.org  Wed Oct 15 22:25:31 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 21:25:31 -0500
Subject: [Wtr-general] Syntax
Message-ID: <3F8E019B.9090306@clabs.org>


> As I've read Paul so far, he's not talking two code bases. He said he
> was thinking of putting his style syntax (tester-friendly) on top of the
> programmer friendly code, so that his is just a wrapper on it. That's
> how I've taken it.
I see - so I was confused. I agree with having tester-friendly code, but I
think you do too - right? The issue is just a matter of deciding on syntax?









From chrismo at clabs.org  Wed Oct 15 22:26:06 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 21:26:06 -0500
Subject: [Wtr-general] RE: WTR dev
Message-ID: <3F8E01BE.9060202@clabs.org>


One challenge I see with doing a generic solution is the differences in tag
declarations people use on the web. For hyperlinks, the only thing I can see
to key on would be the contents href="" portion and perhaps the text between
that and the closing </a>. One really difficult area to look at is things
like radio buttons and drop-downs.

This is a site that I'm going to be working on automating:
http://c0dsp.westjet.com/internet/sky/
If you look at the radio buttons, they are not easy to distinguish between:
<td align=left>
<input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 0,
false);setReturningDayDefault()" checked>
<font size=-2>Round Trip &nbsp;
<input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 1, false )">One Way</font></td>
</td>

You'll notice that each radio button's "name" attribute is identical, so I
would need to key on the name combined with the contents of the "onClick"
attribute. In other instances, I may have identical names, but other tag
attributes may be different.

How do we decide what parameters to pass a function or method that handles a
radio button? If we have "name" parameter but the developers didn't use
"name" attributes, what do we do?






From chrismo at clabs.org  Wed Oct 15 22:36:16 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 21:36:16 -0500
Subject: [Wtr-general] List Replies
Message-ID: <3F8E0420.5090501@clabs.org>

Don't know about anyone else, but I'm still getting weird list behavior 
regarding replies.

Jonathan, I just noticed, your posts tonight have no Reply-To header, so 
my replies default to you direct, not the list. Have you tweaked your 
personal settings on the list in relation to this?

-- 
Chris
http://clabs.org/blogki





From jkohl at telusplanet.net  Wed Oct 15 22:52:06 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Wed, 15 Oct 2003 20:52:06 -0600
Subject: [Wtr-general] List Replies
In-Reply-To: <3F8E0420.5090501@clabs.org>
Message-ID: <000001c39390$7cc40260$6401a8c0@tintin>


> Jonathan, I just noticed, your posts tonight have no Reply-To header, so
> my replies default to you direct, not the list. Have you tweaked your
> personal settings on the list in relation to this?
Not knowingly.




From chrismo at clabs.org  Wed Oct 15 23:05:51 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 22:05:51 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8E019B.9090306@clabs.org>
References: <3F8E019B.9090306@clabs.org>
Message-ID: <3F8E0B0F.9060004@clabs.org>

Jonathan wrote:

> I see - so I was confused. I agree with having tester-friendly code, 
> but I
> think you do too - right? The issue is just a matter of deciding on 
> syntax? 

Yeah.

-- 
Chris
http://clabs.org/blogki





From chrismo at clabs.org  Wed Oct 15 23:06:35 2003
From: chrismo at clabs.org (Chris Morris)
Date: Wed, 15 Oct 2003 22:06:35 -0500
Subject: [Wtr-general] RE: WTR dev
In-Reply-To: <3F8E01BE.9060202@clabs.org>
References: <3F8E01BE.9060202@clabs.org>
Message-ID: <3F8E0B3B.7050708@clabs.org>

Jonathan wrote this post - I forwarded it to the list ...

>
> One challenge I see with doing a generic solution is the differences 
> in tag
> declarations people use on the web. For hyperlinks, ...

--
Chris



From paul.rogers at shaw.ca  Wed Oct 15 23:08:38 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Wed, 15 Oct 2003 21:08:38 -0600
Subject: [Wtr-general] RE: WTR dev
In-Reply-To: <3F8E01BE.9060202@clabs.org>
Message-ID: <000601c39392$cbad5af0$7aa09044@NewDell>

Visual Test does something like this:

clickButton ("xButton")

Where x is something like: + click the button that has a caption Button
                           @ clcik the button that has an ID called
button

I forget exactly, but you get the idea - it uses different symbols to
determine how to identify the button.

In a conversation with Jonathan, I said I wished that I had taken this
approach - the approach I had taken was to use different function names
- clickButtonWithCaption() clickButtonWithName, clickButtonWithID etc

Jonathan seemed to like this way of doing it  - he felt t made the code
more readable.


For the href example below, I ended up writing lots of different
functions - all similar - clickLinkWithURL clickLinkWithText
clickLinkWithID clickLinkWIthURLAndText etc

If you use a regular expression clickLinkWithText( /Logout/ ) you can
get lots of versatility in recognizing the links. You can also get the
wrong ones too :-(

So changing the subject slightly...

The big difference between cLabs Iecontroller and the one I wrote is in
the level of abstraction of the IE functions. In Chriss code, you access
IE almost directly. With my one there is a layer between the user and
IE. This has advantages and disadvantages. The advantages are that other
code can be added into the function - for logging etc. It also makes a
simpler interface for the end user.

The disadvantage is that doing something for which there is no function
is harder.

This also comes back to the developer/qa guy debate. 

I do like the idea of using clabs as the basis with a tester friendly
layer on top. I'll see what I can do along these lines.


Paul



-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Chris Morris
Sent: 15 October 2003 19:26
To: wtr-general at rubyforge.org
Subject: [Wtr-general] RE: WTR dev



One challenge I see with doing a generic solution is the differences in
tag declarations people use on the web. For hyperlinks, the only thing I
can see to key on would be the contents href="" portion and perhaps the
text between that and the closing </a>. One really difficult area to
look at is things like radio buttons and drop-downs.

This is a site that I'm going to be working on automating:
http://c0dsp.westjet.com/internet/sky/
If you look at the radio buttons, they are not easy to distinguish
between: <td align=left> <input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 0,
false);setReturningDayDefault()" checked> <font size=-2>Round Trip
&nbsp; <input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 1, false )">One
Way</font></td> </td>

You'll notice that each radio button's "name" attribute is identical, so
I would need to key on the name combined with the contents of the
"onClick" attribute. In other instances, I may have identical names, but
other tag attributes may be different.

How do we decide what parameters to pass a function or method that
handles a radio button? If we have "name" parameter but the developers
didn't use "name" attributes, what do we do?




_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From paul.rogers at shaw.ca  Wed Oct 15 23:22:53 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Wed, 15 Oct 2003 21:22:53 -0600
Subject: [Wtr-general] RE: WTR dev
In-Reply-To: <000601c39392$cbad5af0$7aa09044@NewDell>
Message-ID: <000701c39394$ca48d700$7aa09044@NewDell>

I think there are several ways of doing the radio button example - the
first is to use a reg expression based on the javascript

clickRadioWithEvent ('travel_type' , 'onClick' , /2/ ) the idea being it
would click the radio button called travel_type, that has an onclick
event that matches a 2

The second way which I prefer, is to use a reg exp to capture the text
before or after the radio, and use that to click the radio ( I know, it
wont work when then the web site is translated...)

clickRadioWithText ( /Round Trip/ , TEXTAFTER ) which would find the
radio button that matched the reg exp 

There are some methods for IE that do similar things - getAdjacentText -
http://msdn.microsoft.com/workshop/author/dhtml/reference/methods/getadj
acenttext.asp?frame=true


Paul

-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Paul Rogers
Sent: 15 October 2003 20:09
To: chrismo at clabs.org; wtr-general at rubyforge.org
Subject: RE: [Wtr-general] RE: WTR dev


Visual Test does something like this:

clickButton ("xButton")

Where x is something like: + click the button that has a caption Button
                           @ clcik the button that has an ID called
button

I forget exactly, but you get the idea - it uses different symbols to
determine how to identify the button.

In a conversation with Jonathan, I said I wished that I had taken this
approach - the approach I had taken was to use different function names
- clickButtonWithCaption() clickButtonWithName, clickButtonWithID etc

Jonathan seemed to like this way of doing it  - he felt t made the code
more readable.


For the href example below, I ended up writing lots of different
functions - all similar - clickLinkWithURL clickLinkWithText
clickLinkWithID clickLinkWIthURLAndText etc

If you use a regular expression clickLinkWithText( /Logout/ ) you can
get lots of versatility in recognizing the links. You can also get the
wrong ones too :-(

So changing the subject slightly...

The big difference between cLabs Iecontroller and the one I wrote is in
the level of abstraction of the IE functions. In Chriss code, you access
IE almost directly. With my one there is a layer between the user and
IE. This has advantages and disadvantages. The advantages are that other
code can be added into the function - for logging etc. It also makes a
simpler interface for the end user.

The disadvantage is that doing something for which there is no function
is harder.

This also comes back to the developer/qa guy debate. 

I do like the idea of using clabs as the basis with a tester friendly
layer on top. I'll see what I can do along these lines.


Paul



-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Chris Morris
Sent: 15 October 2003 19:26
To: wtr-general at rubyforge.org
Subject: [Wtr-general] RE: WTR dev



One challenge I see with doing a generic solution is the differences in
tag declarations people use on the web. For hyperlinks, the only thing I
can see to key on would be the contents href="" portion and perhaps the
text between that and the closing </a>. One really difficult area to
look at is things like radio buttons and drop-downs.

This is a site that I'm going to be working on automating:
http://c0dsp.westjet.com/internet/sky/
If you look at the radio buttons, they are not easy to distinguish
between: <td align=left> <input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 0,
false);setReturningDayDefault()" checked> <font size=-2>Round Trip
&nbsp; <input type=radio name='travel_type'
onClick="combined(0);change_sector_2_fields( 1, false )">One
Way</font></td> </td>

You'll notice that each radio button's "name" attribute is identical, so
I would need to key on the name combined with the contents of the
"onClick" attribute. In other instances, I may have identical names, but
other tag attributes may be different.

How do we decide what parameters to pass a function or method that
handles a radio button? If we have "name" parameter but the developers
didn't use "name" attributes, what do we do?




_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general

_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From jkohl at telusplanet.net  Wed Oct 15 23:31:44 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Wed, 15 Oct 2003 21:31:44 -0600
Subject: [Wtr-general] RE: WTR dev
In-Reply-To: <000601c39392$cbad5af0$7aa09044@NewDell>
Message-ID: <000001c39396$05afa340$6401a8c0@tintin>

> Visual Test does something like this:
> clickButton ("xButton")
> Where x is something like: + click the button that has a caption Button
>                            @ clcik the button that has an ID called
> button
> I forget exactly, but you get the idea - it uses different symbols to
> determine how to identify the button.
> In a conversation with Jonathan, I said I wished that I had taken this
> approach - the approach I had taken was to use different function names
> - clickButtonWithCaption() clickButtonWithName, clickButtonWithID etc
> Jonathan seemed to like this way of doing it  - he felt t made the code
> more readable.
I do feel that the latter approach is more intuitive when writing test
cases. i.e. clickButtonWithCaption() vs. clickButton("+Button")
I did in fact say that I'd rather have different function names that contain
similar code as it makes the test cases easier to follow. 

On further thought, following the SilkTest-style syntax may not work as well
when dealing with a plethora of potential tag attributes vs. an object
captured and stored in a Windows Declaration file on which you use functions
like Click(). Knowing what to click on may dictate what function you use
instead of more generic functions that would work on objects with known
stored properties. In most cases we would be at the mercy of the developer
who could be doing all kinds of things in the tags using different
attributes, naming each element with the same name, and all sorts of other
things that could frustrate the tester's efforts. Since there are a lot of
unknowns when developers declare HTML widgets, multiple combinations of
functions to handle these seems to make sense. I don't know how to deal with
these situations in a generic way as I posted earlier. 


 > For the href example below, I ended up writing lots of different
> functions - all similar - clickLinkWithURL clickLinkWithText
> clickLinkWithID clickLinkWIthURLAndText etc
This is where the differences between tag attributes being used would be
addressed. That's where I struggle with a generic solution vs. a lot of
similar combinations I think. That's why I posed that problem to the group.
I understand where Paul is coming from on this design, but that's easier for
me because he demonstrated it for me.

I think the syntactic issue is less about a particular style, and more to do
with handling these unique situations that you get when dealing with tags
that have different combinations of attributes. Is that correct Paul?


> So changing the subject slightly...
> The big difference between cLabs Iecontroller and the one I wrote is in
> the level of abstraction of the IE functions. In Chriss code, you access
> IE almost directly. With my one there is a layer between the user and
> IE. This has advantages and disadvantages. The advantages are that other
> code can be added into the function - for logging etc. It also makes a
> simpler interface for the end user.
> The disadvantage is that doing something for which there is no function
> is harder.
> This also comes back to the developer/qa guy debate.
> I do like the idea of using clabs as the basis with a tester friendly
> layer on top. I'll see what I can do along these lines.
That sounds interesting. As one of the Ruby Nuby's here, getting a grasp on
the development direction is more of a challenge. I think I have a better
handle on what the issues are after today's exchange. So we would
essentially have two components with this model:
1) core Clabs Controller				(engine)
2) tester layer that drives the controller 	(tester interface)

So some could just drive from the Clabs Controller, and others could drive
from the "tester development layer". Using either method, using Test::Unit
as the harness would be very beneficial.









From bret at pettichord.com  Thu Oct 16 01:43:04 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:43:04 -0500
Subject: [Wtr-general] Paul's code
In-Reply-To: <c5e03c597b.c597bc5e03@shaw.ca>
Message-ID: <5.1.0.14.2.20031016003800.01c65b00@127.0.0.1>

At 01:59 PM 10/15/2003, paul.rogers at shaw.ca wrote:
>Ideally I would like to add this as a layer on top of the cLabs code so we 
>have just the single code base, although Im not too sure how this would 
>work out right now.

This is also my thinking. We have two different ways of driving the code, 
but one is a layer on top of the other. Some people may like one over the 
other. If one of the ways doesn't have many advantages, it will wither away.

Right now i'd like to see everything checked in, perhaps as two different 
controllers.

What i would really like to see are examples that exercise each type of 
interface. Then we can refactor and merge the code and make sure that each 
still works.

I can see that at some point we may have reason to choose one way or 
another, but i don't think we are close to that yet.

BTW -- Brian and i taught our class on monday and i also demoed 
iecontroller today here in Portland at PNSQC. I'm seeing a lot of interest 
in this and have a list of stuff that i need to make happen in the next 
week or two, before we teach it again in San Jose.

I'm am very excited about this. I have a bunch of horribly organized code 
that i need to tidy up and share with you all. I appreciate your patience.

Bret


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 01:47:38 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:47:38 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8D9F70.2010802@clabs.org>
References: <3F857F95.1020308@clabs.org>
 <3F857F95.1020308@clabs.org>
Message-ID: <5.1.0.14.2.20031016004409.01f51d18@127.0.0.1>

At 02:26 PM 10/15/2003, Chris Morris wrote:
>I fear an impasse here. Syntax style is not going to be proven 
>definitively easier one way or the other. Obviously, whatever one's own 
>prior experience is will greatly influence that person's definition of 
>easier. Right now there's a thread on the Ruby list about the readability 
>of parentheses around parameters to a method call. It's getting equal hits 
>from across the spectrum.
>
>My concern with having two different styles is that it will make the 
>library itself more unpredictable to the user. Here we do this, there we 
>do that -- unless we try to offer two layers as Paul suggested. Then my 
>concern is with coding time -- I'm not going to have much time to 
>contribute, everything's going to be volunteer, and trying to maintain two 
>styles at the same time sounds like a bit much.
>Thoughts?

I say lets encourage diversity for now. In the worst case, we'll have two 
controllers and perhaps have to split them into two projects. But lets put 
them together right now.

I would really like to do some remote pairing with others while really 
working on this code. Is anyone else set up to do this? One way would 
simply be to create a branch in cvs to work together on and then use IM or 
a phone to work on it together. WDYT?



>--
>Chris
>http://clabs.org/blogki
>
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 01:54:27 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:54:27 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8DDE60.70404@clabs.org>
References: <000001c39372$085aa960$6401a8c0@tintin>
	<000001c39372$085aa960$6401a8c0@tintin>
Message-ID: <5.1.0.14.2.20031016005005.01f099b8@127.0.0.1>

At 06:55 PM 10/15/2003, Chris Morris wrote:
>btw, I do agree with the Test::Unit usage. That framework buys us a lot of 
>small stuff we won't have to hassle with, and it's standard in the Ruby 
>community.

I agree. I'm also in favor of methods as

form.address.assert_value("Main St")

You can do this by mixing in assertions. You also have to do this, to keep 
assertion failures from killing IRB. Ultimately, this really should be a 
change to Test::Unit. Remind to send this to Nathanial. I suspect this was 
an oversite, but their could be a reason for the existing parentage of 
AssertionFailedError:

require 'test/unit/assertions'
module Test
   module Unit
     remove_const(:AssertionFailedError) # to avoid warning
     class AssertionFailedError < RuntimeError
     end
   end
end
include Test::Unit::Assertions





_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 01:54:27 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:54:27 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8DDE60.70404@clabs.org>
References: <000001c39372$085aa960$6401a8c0@tintin>
	<000001c39372$085aa960$6401a8c0@tintin>
Message-ID: <5.1.0.14.2.20031016005005.01f099b8@127.0.0.1>

At 06:55 PM 10/15/2003, Chris Morris wrote:
>btw, I do agree with the Test::Unit usage. That framework buys us a lot of 
>small stuff we won't have to hassle with, and it's standard in the Ruby 
>community.

I agree. I'm also in favor of methods as

form.address.assert_value("Main St")

You can do this by mixing in assertions. You also have to do this, to keep 
assertion failures from killing IRB. Ultimately, this really should be a 
change to Test::Unit. Remind to send this to Nathanial. I suspect this was 
an oversite, but their could be a reason for the existing parentage of 
AssertionFailedError:

require 'test/unit/assertions'
module Test
   module Unit
     remove_const(:AssertionFailedError) # to avoid warning
     class AssertionFailedError < RuntimeError
     end
   end
end
include Test::Unit::Assertions





_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 01:55:32 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:55:32 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <3F8DDE60.70404@clabs.org>
References: <000001c39372$085aa960$6401a8c0@tintin>
	<000001c39372$085aa960$6401a8c0@tintin>
Message-ID: <5.1.0.14.2.20031016005446.01c60260@127.0.0.1>

At 06:55 PM 10/15/2003, you wrote:
>This is the tricky bit, defining what's readable and easy to use to a 
>large group of people.
>
>Perhaps I can take the Google snippet from before and work on keeping the 
>syntax style I prefer and improving the readability for the non-developer.

Right, let's create examples of each of the styles and then solicit 
comments on the wiki. Then we'll find out what people/testers like and why.



_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 01:57:02 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 00:57:02 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <000001c39379$64c23180$6401a8c0@tintin>
References: <3F8DDE60.70404@clabs.org>
Message-ID: <5.1.0.14.2.20031016005621.01f5f9b8@127.0.0.1>

At 07:06 PM 10/15/2003, Jonathan Kohl wrote:
>So do we need to choose a style, agree and move forward at this point? Any
>other thoughts with regard to syntax styles? How do we address these
>concerns and move forward?

If this remains an issue of concern, we should choose a time and IM on it. 
i don't think email is really working.



_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Thu Oct 16 16:40:51 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Thu, 16 Oct 2003 15:40:51 -0500
Subject: [Wtr-general] RE: WTR dev
In-Reply-To: <000601c39392$cbad5af0$7aa09044@NewDell>
References: <3F8E01BE.9060202@clabs.org>
Message-ID: <5.1.0.14.2.20031016153723.01e8eca8@127.0.0.1>

What about something like this:

clickButton (:Caption "foo")
clickButton (:Name "foo")
clickButton (:ID "foo")


At 10:08 PM 10/15/2003, Paul Rogers wrote:
>Visual Test does something like this:
>
>clickButton ("xButton")
>
>Where x is something like: + click the button that has a caption Button
>                            @ clcik the button that has an ID called
>button
>
>I forget exactly, but you get the idea - it uses different symbols to
>determine how to identify the button.
>
>In a conversation with Jonathan, I said I wished that I had taken this
>approach - the approach I had taken was to use different function names
>- clickButtonWithCaption() clickButtonWithName, clickButtonWithID etc
>
>Jonathan seemed to like this way of doing it  - he felt t made the code
>more readable.

I have doc for both SilkTest and VisualTest. I'll try to put their syntaxes 
on the wiki.

Bret

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Thu Oct 16 20:49:57 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 16 Oct 2003 19:49:57 -0500
Subject: [Wtr-general] Syntax
In-Reply-To: <5.1.0.14.2.20031016005621.01f5f9b8@127.0.0.1>
References: <3F8DDE60.70404@clabs.org>
	<5.1.0.14.2.20031016005621.01f5f9b8@127.0.0.1>
Message-ID: <3F8F3CB5.8030907@clabs.org>

Bret Pettichord wrote:

> At 07:06 PM 10/15/2003, Jonathan Kohl wrote:
>
>> So do we need to choose a style, agree and move forward at this 
>> point? Any
>> other thoughts with regard to syntax styles? How do we address these
>> concerns and move forward?
>
>
> If this remains an issue of concern, we should choose a time and IM on 
> it. i don't think email is really working. 


Yeah, probably time to start talking code.

-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Thu Oct 16 21:00:23 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 16 Oct 2003 20:00:23 -0500
Subject: [Wtr-general] [Fwd: Re: Mozilla ActiveX Control installation]
Message-ID: <3F8F3F27.2050902@clabs.org>

Well, bad news on the Mozilla front. I was right - it's just a control 
and cannot work be instantiated by itself.

One last hope at the moment -- IE can host the Mozilla control itself -- 
so maybe we can drive IE hosting the Mozilla control and somehow make 
that work -- but ... I dunno.

Adam Lock is the author of the MozAxc.

-------- Original Message --------
Subject: 	Re: Mozilla ActiveX Control installation
Date: 	Thu, 16 Oct 2003 19:15:26 +0000
From: 	Adam Lock <adamlock at eircom.net>
To: 	chrismo at clabs.org
References: 	<3F86BEB3.9000600 at clabs.org> <3F87C6CE.1050106 at eircom.net> 
<3F8AA91A.8040600 at clabs.org>



Chris Morris wrote:

>
> Here's the little Ruby script:
>
>    require 'win32ole'
>
>    @moz = WIN32OLE.new('Mozilla.Browser.1') # Mozilla.Browser does not 
> work
>    @moz.gohome
>
Hi

You can't do this (not even with the IE control). Controls are tricky 
things to manage and you normally have to write a lot of code yourself, 
or use a language such as VB or C++ which has classes that do it for 
you. Generally you need to create a COM object called a control site 
that implements various COM interfaces and manages a control object - 
activating the control, sizing it and so on.

The approach you have above would only work for OLE automation objects 
which might include the likes of MS Word and possibly the IE browser 
*application*, but not the Mozilla or IE controls. The controls are DLLs 
which live in a window, not an executable that you can just create an 
instance of and remotely control from somewhere else.

Adam


-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Thu Oct 16 22:42:24 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 16 Oct 2003 21:42:24 -0500
Subject: [Wtr-general] [Fwd: Re: Mozilla ActiveX Control installation]
In-Reply-To: <000401c39455$42c76280$7aa09044@NewDell>
References: <000401c39455$42c76280$7aa09044@NewDell>
Message-ID: <3F8F5710.40202@clabs.org>

Paul Rogers wrote:

>How about this...
>
>
>Use a vb form to host the mozilla control
>

Yeah, I'd thought about that -- not much difference tween that and 
hosting it inside IE -- both are acting as a container. We'd have to 
dink with it.

-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Fri Oct 17 09:09:45 2003
From: chrismo at clabs.org (Chris Morris)
Date: Fri, 17 Oct 2003 08:09:45 -0500
Subject: [Wtr-general] Paul's code
In-Reply-To: <cb09fc6ff5.c6ff5cb09f@shaw.ca>
References: <cb09fc6ff5.c6ff5cb09f@shaw.ca>
Message-ID: <3F8FEA19.6090406@clabs.org>

paul.rogers at shaw.ca wrote:

Paul's code has been added to the CVS repository. I put it under 
wtr/contrib/paulr because I didn't have time to look it over and see if 
there was a better location for it.

-- 
Chris
http://clabs.org/blogki




From paul.rogers at shaw.ca  Sun Oct 19 13:38:55 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sun, 19 Oct 2003 11:38:55 -0600
Subject: [Wtr-general] Another version of my controller
Message-ID: <000001c39667$deb33430$7aa09044@NewDell>

I still havent got cvs working, so here is another zip of the code so
far.

This adds select box and link clicking.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ieController.zip
Type: application/x-zip-compressed
Size: 3294 bytes
Desc: not available
Url : http://rubyforge.org/pipermail/wtr-general/attachments/20031019/919fc608/attachment.bin 

From chrismo at clabs.org  Sun Oct 19 18:08:51 2003
From: chrismo at clabs.org (Chris Morris)
Date: Sun, 19 Oct 2003 17:08:51 -0500
Subject: [Wtr-general] Driving Mozilla
Message-ID: <3F930B73.2060403@clabs.org>

 From the MozAxc author:

> (To give some context to this thread, a few of us have been working 
> with Ruby to drive IE for automated testing of web apps, and would 
> love to have the same scripts drive Mozilla as well to test it out. 
> http://clabs.org/wtr)
>
Hi,

You can automate Mozilla but not through ActiveX. The whole frontend is 
just one giant javascript programme with XML so there are numerous ways 
you could hook into it, either by editing the source of the UI directly, 
or by writing JS that has additional privileges, for example to grab the 
nsIWebNavigation interface of another window and drive it much like you 
do in IE. It's worth asking in the netscape.public.mozilla.xpfe groups 
about this kind of thing.

You might also like to look at the browser buster on Mozilla.org.

http://www.mozilla.org/quality/browser/buster/

This site demonstrates how to automate Mozilla through the top 100 
websites for regression testing. The script goes in one frame and drives 
the content in the other. You should get the idea when you see it.

Hope this helps

Adam

-- 
Chris
http://clabs.org/blogki





From groupmember at peterw.nl  Fri Oct 24 10:13:02 2003
From: groupmember at peterw.nl (groupmember at peterw.nl)
Date: Fri, 24 Oct 2003 16:13:02 +0200
Subject: [Wtr-general] Getting at the right form
Message-ID: <00b301c39a38$f00ff6d0$2f02a8c0@PWLaptop>




I have an application that I want to test with ruby and cliecontroller.rb 
the thing is that the url to test is always creating a popup window and then closes the original IE window. 
I would like to get to the form property of the newly created window but I don't know how to get this done with the COM model 
 
anyone?

ie = ClIEController.new true
 
ie.navigate "www.xxx.yyy.com/url"
ie.wait #wait till ie has loaded the page and the form is present
puts "FORMS" + ie.document.forms.length.to_s
forms = ie.document.forms 
for i in 1...forms.length
        puts forms[i].name
end

there a no forms to show, since I am (probably) referring to the wrong document.

thanks very much!

regards Peter


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20031024/a6fc3582/attachment.html 

From groupmember at peterw.nl  Fri Oct 24 10:30:12 2003
From: groupmember at peterw.nl (groupmember at peterw.nl)
Date: Fri, 24 Oct 2003 16:30:12 +0200
Subject: [Wtr-general] Getting at the right form
Message-ID: <001d01c39a3b$55f78a60$2f02a8c0@PWLaptop>

I have an application that I want to test with ruby and cliecontroller.rb 
the thing is that the url to test is always creating a popup window and then closes the original IE window. 
I would like to get to the form property of the newly created window but I don't know how to get this done with the COM model 
 
anyone?

ie = ClIEController.new true
 
ie.navigate "www.xxx.yyy.com/url"
ie.wait #wait till ie has loaded the page and the form is present
puts "FORMS" + ie.document.forms.length.to_s
forms = ie.document.forms 
for i in 1...forms.length
        puts forms[i].name
end

there a no forms to show, since I am (probably) referring to the wrong document.

thanks very much!

regards Peter


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20031024/fecace0a/attachment.html 

From paul.rogers at shaw.ca  Sat Oct 25 10:12:50 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 08:12:50 -0600
Subject: [Wtr-general] Getting at the right form
In-Reply-To: <001d01c39a3b$55f78a60$2f02a8c0@PWLaptop>
Message-ID: <001101c39b02$12d00b50$7aa09044@NewDell>

Chris would be the best to answer this one. But I think there is a bug
in Ruby ( Im still writing the test code to demonstrate it) where the
Win32OLE component returns an incorrect data type when a window opens.
This means you cant get an OLE reference to the new window.
 
Paul

-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of
groupmember at peterw.nl
Sent: 24 October 2003 07:30
To: wtr-general at rubyforge.org
Subject: [Wtr-general] Getting at the right form


I have an application that I want to test with ruby and
cliecontroller.rb 
the thing is that the url to test is always creating a popup window and
then closes the original IE window. 
I would like to get to the form property of the newly created window but
I don't know how to get this done with the COM model 
 
anyone?
 

ie = ClIEController.new true
 
ie.navigate "www.xxx.yyy.com/url"
ie.wait #wait till ie has loaded the page and the form is present
puts "FORMS" + ie.document.forms.length.to_s
forms = ie.document.forms 
for i in 1...forms.length
        puts forms[i].name
end
 
there a no forms to show, since I am (probably) referring to the wrong
document.
 
thanks very much!
 
regards Peter



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20031025/1bf38b19/attachment.html 

From paul.rogers at shaw.ca  Sat Oct 25 14:37:55 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 12:37:55 -0600
Subject: [Wtr-general] Iecontroller with exceptions
Message-ID: <002001c39b27$1aec00d0$7aa09044@NewDell>

Ive attached here a zip of my iecontroller that throws exceptions
instead of returning messages, and a test page using test::unit

Its probably my inexperience of using test::unit , but I don't think
this provides as useful a way of testing web apps as the previous
version. ( which returned  an array of strings detaling the problem)

Heres why:
If I try to click a button that isnt there, I need to catch several
possible exceptions - disable, missing etc. Now with test::unit I don't
seem to be able to catch an exception unless I know what the exception
is. The point here is that when testing an application, I don't know
what exception will be thrown - unlike unit testing where I am making
sure the exception happens.

This makes it a little harder to act on a problem with a button (or
other object).

Paul



From paul.rogers at shaw.ca  Sat Oct 25 14:38:56 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 12:38:56 -0600
Subject: [Wtr-general] RE: Iecontroller with exceptions
Message-ID: <002101c39b27$3f373b80$7aa09044@NewDell>

This time with the zip



-----Original Message-----
From: Paul Rogers [mailto:paul.rogers at shaw.ca] 
Sent: 25 October 2003 11:38
To: wtr-general at rubyforge.org
Subject: Iecontroller with exceptions


Ive attached here a zip of my iecontroller that throws exceptions
instead of returning messages, and a test page using test::unit

Its probably my inexperience of using test::unit , but I don't think
this provides as useful a way of testing web apps as the previous
version. ( which returned  an array of strings detaling the problem)

Heres why:
If I try to click a button that isnt there, I need to catch several
possible exceptions - disable, missing etc. Now with test::unit I don't
seem to be able to catch an exception unless I know what the exception
is. The point here is that when testing an application, I don't know
what exception will be thrown - unlike unit testing where I am making
sure the exception happens.

This makes it a little harder to act on a problem with a button (or
other object).

Paul
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ieController_exception.zip
Type: application/x-zip-compressed
Size: 3504 bytes
Desc: not available
Url : http://rubyforge.org/pipermail/wtr-general/attachments/20031025/778e2b6c/attachment.bin 

From jkohl at telusplanet.net  Sat Oct 25 14:44:40 2003
From: jkohl at telusplanet.net (Jonathan Kohl)
Date: Sat, 25 Oct 2003 12:44:40 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <002001c39b27$1aec00d0$7aa09044@NewDell>
Message-ID: <000001c39b28$0c89cb20$6401a8c0@tintin>

Hey Paul;

So you feel that using Test::Unit isn't a good match for the Controller
you've developed? I was hoping to use Test::Unit in our continuous
integration environment using Ant and Anthill for builds, and JUnit,
HTTPUnit and the WTR IEController. That way we would have the code coverage
with JUnit, the HTTP response testing with HTTPUnit and the user interface
tests using the WTR IEController.  We are using a couple other tools to
measure and report coverage such as JDepend and JCoverage, so we might be
able to get around using Test::Unit in the continuous integration process if
it doesn't work out.

I did a proof of concept yesterday using the latest version of the
Controller that you sent out, and the developers were thrilled with what
could be done with the WTR Controller. 

-Jonathan

> -----Original Message-----
> From: wtr-general-bounces at rubyforge.org [mailto:wtr-general-
> bounces at rubyforge.org] On Behalf Of Paul Rogers
> Sent: October 25, 2003 12:38 PM
> To: wtr-general at rubyforge.org
> Subject: [Wtr-general] Iecontroller with exceptions
> 
> Ive attached here a zip of my iecontroller that throws exceptions
> instead of returning messages, and a test page using test::unit
> 
> Its probably my inexperience of using test::unit , but I don't think
> this provides as useful a way of testing web apps as the previous
> version. ( which returned  an array of strings detaling the problem)
> 
> Heres why:
> If I try to click a button that isnt there, I need to catch several
> possible exceptions - disable, missing etc. Now with test::unit I don't
> seem to be able to catch an exception unless I know what the exception
> is. The point here is that when testing an application, I don't know
> what exception will be thrown - unlike unit testing where I am making
> sure the exception happens.
> 
> This makes it a little harder to act on a problem with a button (or
> other object).
> 
> Paul
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general





From bret at pettichord.com  Sat Oct 25 14:50:30 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sat, 25 Oct 2003 13:50:30 -0500
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <002001c39b27$1aec00d0$7aa09044@NewDell>
Message-ID: <5.1.0.14.2.20031025134816.01f07a08@127.0.0.1>

Paul,

If you catch RuntimeError, you'll also catch anything that derives 
(inherits) from it. That should do what you want. If not, try catching 
StandardError or even Exception. (See page 93 of the printed pickaxe book 
-- this diagram isn't in the online version.)

Bret

At 01:37 PM 10/25/2003, Paul Rogers wrote:
>Ive attached here a zip of my iecontroller that throws exceptions
>instead of returning messages, and a test page using test::unit
>
>Its probably my inexperience of using test::unit , but I don't think
>this provides as useful a way of testing web apps as the previous
>version. ( which returned  an array of strings detaling the problem)
>
>Heres why:
>If I try to click a button that isnt there, I need to catch several
>possible exceptions - disable, missing etc. Now with test::unit I don't
>seem to be able to catch an exception unless I know what the exception
>is. The point here is that when testing an application, I don't know
>what exception will be thrown - unlike unit testing where I am making
>sure the exception happens.
>
>This makes it a little harder to act on a problem with a button (or
>other object).
>
>Paul
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From bret at pettichord.com  Sat Oct 25 14:50:30 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sat, 25 Oct 2003 13:50:30 -0500
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <002001c39b27$1aec00d0$7aa09044@NewDell>
Message-ID: <5.1.0.14.2.20031025134816.01f07a08@127.0.0.1>

Paul,

If you catch RuntimeError, you'll also catch anything that derives 
(inherits) from it. That should do what you want. If not, try catching 
StandardError or even Exception. (See page 93 of the printed pickaxe book 
-- this diagram isn't in the online version.)

Bret

At 01:37 PM 10/25/2003, Paul Rogers wrote:
>Ive attached here a zip of my iecontroller that throws exceptions
>instead of returning messages, and a test page using test::unit
>
>Its probably my inexperience of using test::unit , but I don't think
>this provides as useful a way of testing web apps as the previous
>version. ( which returned  an array of strings detaling the problem)
>
>Heres why:
>If I try to click a button that isnt there, I need to catch several
>possible exceptions - disable, missing etc. Now with test::unit I don't
>seem to be able to catch an exception unless I know what the exception
>is. The point here is that when testing an application, I don't know
>what exception will be thrown - unlike unit testing where I am making
>sure the exception happens.
>
>This makes it a little harder to act on a problem with a button (or
>other object).
>
>Paul
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From paul.rogers at shaw.ca  Sat Oct 25 15:11:02 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 13:11:02 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <000001c39b28$0c89cb20$6401a8c0@tintin>
Message-ID: <002b01c39b2b$bb503240$7aa09044@NewDell>

I also did a proof of concept for my new employer yesterday. The outcome
was that we would ditch Silk in favour of ruby. Various people were
pleased we no longer had to pay the Silk license fee, and I was pleased
I didn't have to go back to using it.

I have an updated controller at work, which I hope to be able to send
out next week.

Bret has just sent out a reply with some detail on how I could rework
the exceptions, so I'll try that now

Paul



-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Jonathan Kohl
Sent: 25 October 2003 11:45
To: wtr-general at rubyforge.org
Subject: RE: [Wtr-general] Iecontroller with exceptions


Hey Paul;

So you feel that using Test::Unit isn't a good match for the Controller
you've developed? I was hoping to use Test::Unit in our continuous
integration environment using Ant and Anthill for builds, and JUnit,
HTTPUnit and the WTR IEController. That way we would have the code
coverage with JUnit, the HTTP response testing with HTTPUnit and the
user interface tests using the WTR IEController.  We are using a couple
other tools to measure and report coverage such as JDepend and
JCoverage, so we might be able to get around using Test::Unit in the
continuous integration process if it doesn't work out.

I did a proof of concept yesterday using the latest version of the
Controller that you sent out, and the developers were thrilled with what
could be done with the WTR Controller. 

-Jonathan

> -----Original Message-----
> From: wtr-general-bounces at rubyforge.org [mailto:wtr-general- 
> bounces at rubyforge.org] On Behalf Of Paul Rogers
> Sent: October 25, 2003 12:38 PM
> To: wtr-general at rubyforge.org
> Subject: [Wtr-general] Iecontroller with exceptions
> 
> Ive attached here a zip of my iecontroller that throws exceptions 
> instead of returning messages, and a test page using test::unit
> 
> Its probably my inexperience of using test::unit , but I don't think 
> this provides as useful a way of testing web apps as the previous 
> version. ( which returned  an array of strings detaling the problem)
> 
> Heres why:
> If I try to click a button that isnt there, I need to catch several 
> possible exceptions - disable, missing etc. Now with test::unit I 
> don't seem to be able to catch an exception unless I know what the 
> exception is. The point here is that when testing an application, I 
> don't know what exception will be thrown - unlike unit testing where I

> am making sure the exception happens.
> 
> This makes it a little harder to act on a problem with a button (or 
> other object).
> 
> Paul
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org 
> http://rubyforge.org/mailman/listinfo/wtr-general



_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From paul.rogers at shaw.ca  Sat Oct 25 15:52:12 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 13:52:12 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <5.1.0.14.2.20031025134816.01f07a08@127.0.0.1>
Message-ID: <002c01c39b31$7b9a7a10$7aa09044@NewDell>

It could be that Im so sure my method of passing strings back as
failures that Im not open to the exception method, but Im still not
seeing how using exceptions or test::unit benefits.

When unit testing we are supplying input parameters that will cause an
exception, hence unit::test wworks well.

Whereas with application level testing (where I see this controller
being most beneficial) we don't expect exceptions to occur, and to have
to trap every exception seems like a lot of code that is rarely used.

Am I making sense??

Paul






-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
Sent: 25 October 2003 11:51
To: wtr-general at rubyforge.org; wtr-general at rubyforge.org
Subject: Re: [Wtr-general] Iecontroller with exceptions


Paul,

If you catch RuntimeError, you'll also catch anything that derives 
(inherits) from it. That should do what you want. If not, try catching 
StandardError or even Exception. (See page 93 of the printed pickaxe
book 
-- this diagram isn't in the online version.)

Bret

At 01:37 PM 10/25/2003, Paul Rogers wrote:
>Ive attached here a zip of my iecontroller that throws exceptions 
>instead of returning messages, and a test page using test::unit
>
>Its probably my inexperience of using test::unit , but I don't think 
>this provides as useful a way of testing web apps as the previous 
>version. ( which returned  an array of strings detaling the problem)
>
>Heres why:
>If I try to click a button that isnt there, I need to catch several 
>possible exceptions - disable, missing etc. Now with test::unit I don't

>seem to be able to catch an exception unless I know what the exception 
>is. The point here is that when testing an application, I don't know 
>what exception will be thrown - unlike unit testing where I am making 
>sure the exception happens.
>
>This makes it a little harder to act on a problem with a button (or 
>other object).
>
>Paul
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org 
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com

_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From bret at pettichord.com  Sat Oct 25 16:27:49 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sat, 25 Oct 2003 15:27:49 -0500
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <002c01c39b31$7b9a7a10$7aa09044@NewDell>
References: <5.1.0.14.2.20031025134816.01f07a08@127.0.0.1>
Message-ID: <5.1.0.14.2.20031025152655.01f0eb28@127.0.0.1>

Paul,

I haven't looked at your code, but i don't understand your concern. Doesn't 
Test::unit basically work just like SilkTest as regards exceptions?

Bret

At 02:52 PM 10/25/2003, Paul Rogers wrote:
>It could be that Im so sure my method of passing strings back as
>failures that Im not open to the exception method, but Im still not
>seeing how using exceptions or test::unit benefits.
>
>When unit testing we are supplying input parameters that will cause an
>exception, hence unit::test wworks well.
>
>Whereas with application level testing (where I see this controller
>being most beneficial) we don't expect exceptions to occur, and to have
>to trap every exception seems like a lot of code that is rarely used.
>
>Am I making sense??
>
>Paul
>
>
>
>
>
>
>-----Original Message-----
>From: wtr-general-bounces at rubyforge.org
>[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
>Sent: 25 October 2003 11:51
>To: wtr-general at rubyforge.org; wtr-general at rubyforge.org
>Subject: Re: [Wtr-general] Iecontroller with exceptions
>
>
>Paul,
>
>If you catch RuntimeError, you'll also catch anything that derives
>(inherits) from it. That should do what you want. If not, try catching
>StandardError or even Exception. (See page 93 of the printed pickaxe
>book
>-- this diagram isn't in the online version.)
>
>Bret
>
>At 01:37 PM 10/25/2003, Paul Rogers wrote:
> >Ive attached here a zip of my iecontroller that throws exceptions
> >instead of returning messages, and a test page using test::unit
> >
> >Its probably my inexperience of using test::unit , but I don't think
> >this provides as useful a way of testing web apps as the previous
> >version. ( which returned  an array of strings detaling the problem)
> >
> >Heres why:
> >If I try to click a button that isnt there, I need to catch several
> >possible exceptions - disable, missing etc. Now with test::unit I don't
>
> >seem to be able to catch an exception unless I know what the exception
> >is. The point here is that when testing an application, I don't know
> >what exception will be thrown - unlike unit testing where I am making
> >sure the exception happens.
> >
> >This makes it a little harder to act on a problem with a button (or
> >other object).
> >
> >Paul
> >
> >_______________________________________________
> >Wtr-general mailing list
> >Wtr-general at rubyforge.org
> >http://rubyforge.org/mailman/listinfo/wtr-general
>
>_____________________________________
>   Bret Pettichord, Software Tester
>     Book - www.testinglessons.com
>     Consulting - www.pettichord.com
>     Blog - www.io.com/~wazmo/blog
>     Hotlist - www.testinghotlist.com
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From paul.rogers at shaw.ca  Sat Oct 25 16:55:04 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sat, 25 Oct 2003 14:55:04 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <5.1.0.14.2.20031025152655.01f0eb28@127.0.0.1>
Message-ID: <002d01c39b3a$43df4520$7aa09044@NewDell>

I cant remember what silk does. Id have to look on Monday. Ive given up
for today.


-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
Sent: 25 October 2003 13:28
To: wtr-general at rubyforge.org
Subject: RE: [Wtr-general] Iecontroller with exceptions


Paul,

I haven't looked at your code, but i don't understand your concern.
Doesn't 
Test::unit basically work just like SilkTest as regards exceptions?

Bret

At 02:52 PM 10/25/2003, Paul Rogers wrote:
>It could be that Im so sure my method of passing strings back as 
>failures that Im not open to the exception method, but Im still not 
>seeing how using exceptions or test::unit benefits.
>
>When unit testing we are supplying input parameters that will cause an 
>exception, hence unit::test wworks well.
>
>Whereas with application level testing (where I see this controller 
>being most beneficial) we don't expect exceptions to occur, and to have

>to trap every exception seems like a lot of code that is rarely used.
>
>Am I making sense??
>
>Paul
>
>
>
>
>
>
>-----Original Message-----
>From: wtr-general-bounces at rubyforge.org 
>[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
>Sent: 25 October 2003 11:51
>To: wtr-general at rubyforge.org; wtr-general at rubyforge.org
>Subject: Re: [Wtr-general] Iecontroller with exceptions
>
>
>Paul,
>
>If you catch RuntimeError, you'll also catch anything that derives
>(inherits) from it. That should do what you want. If not, try catching 
>StandardError or even Exception. (See page 93 of the printed pickaxe 
>book
>-- this diagram isn't in the online version.)
>
>Bret
>
>At 01:37 PM 10/25/2003, Paul Rogers wrote:
> >Ive attached here a zip of my iecontroller that throws exceptions 
> >instead of returning messages, and a test page using test::unit
> >
> >Its probably my inexperience of using test::unit , but I don't think 
> >this provides as useful a way of testing web apps as the previous 
> >version. ( which returned  an array of strings detaling the problem)
> >
> >Heres why:
> >If I try to click a button that isnt there, I need to catch several 
> >possible exceptions - disable, missing etc. Now with test::unit I 
> >don't
>
> >seem to be able to catch an exception unless I know what the 
> >exception is. The point here is that when testing an application, I 
> >don't know what exception will be thrown - unlike unit testing where 
> >I am making sure the exception happens.
> >
> >This makes it a little harder to act on a problem with a button (or 
> >other object).
> >
> >Paul
> >
> >_______________________________________________
> >Wtr-general mailing list
> >Wtr-general at rubyforge.org 
> >http://rubyforge.org/mailman/listinfo/wtr-general
>
>_____________________________________
>   Bret Pettichord, Software Tester
>     Book - www.testinglessons.com
>     Consulting - www.pettichord.com
>     Blog - www.io.com/~wazmo/blog
>     Hotlist - www.testinghotlist.com
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org 
>http://rubyforge.org/mailman/listinfo/wtr-general
>
>_______________________________________________
>Wtr-general mailing list
>Wtr-general at rubyforge.org 
>http://rubyforge.org/mailman/listinfo/wtr-general

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com

_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From bret at pettichord.com  Sat Oct 25 18:52:03 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sat, 25 Oct 2003 17:52:03 -0500
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <002d01c39b3a$43df4520$7aa09044@NewDell>
References: <5.1.0.14.2.20031025152655.01f0eb28@127.0.0.1>
Message-ID: <5.1.0.14.2.20031025175059.040b9c98@127.0.0.1>

At 03:55 PM 10/25/2003, Paul Rogers wrote:
>I cant remember what silk does. Id have to look on Monday. Ive given up
>for today.

If you, for example, try to select an item in a control that isn't enable, 
silktest will throw a non-enabled exception.





_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From paul.rogers at shaw.ca  Sun Oct 26 13:07:37 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sun, 26 Oct 2003 11:07:37 -0700
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <5.1.0.14.2.20031025175059.040b9c98@127.0.0.1>
Message-ID: <000101c39bec$099c6f20$7aa09044@NewDell>

That is how Ive done it in the latest controller. One of the big
advantages of returning strings as the error rather than throwing
exceptions, was the ability to do negative testing with virtually no
extra over head. Using exceptions is probably a more correct way of
doing it but seems to me to generate a lot of extra overhead.

I'll continue to look into the possibilites around this over then next
few days.

Paul

-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
Sent: 25 October 2003 15:52
To: wtr-general at rubyforge.org
Subject: RE: [Wtr-general] Iecontroller with exceptions


At 03:55 PM 10/25/2003, Paul Rogers wrote:
>I cant remember what silk does. Id have to look on Monday. Ive given up

>for today.

If you, for example, try to select an item in a control that isn't
enable, 
silktest will throw a non-enabled exception.





_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com

_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From bret at pettichord.com  Sun Oct 26 13:16:08 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Sun, 26 Oct 2003 12:16:08 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <000101c39bec$099c6f20$7aa09044@NewDell>
References: <5.1.0.14.2.20031025175059.040b9c98@127.0.0.1>
Message-ID: <5.1.0.14.2.20031026121318.01ec6dd8@127.0.0.1>

At 12:07 PM 10/26/2003, Paul Rogers wrote:
>That is how Ive done it in the latest controller. One of the big
>advantages of returning strings as the error rather than throwing
>exceptions, was the ability to do negative testing with virtually no
>extra over head. Using exceptions is probably a more correct way of
>doing it but seems to me to generate a lot of extra overhead.

Have you tried using assert_raises?

I believe that signalling errors by return code requires more overhead 
because wrapper functions must be sure to check and return the values 
returned by the functions they call. Not doing this, or not doing this 
correctly, can cause testsuites to ignore triggered errors - a fatal flaw.


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From paul.rogers at shaw.ca  Sun Oct 26 19:57:57 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sun, 26 Oct 2003 17:57:57 -0700
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <5.1.0.14.2.20031026121318.01ec6dd8@127.0.0.1>
Message-ID: <000001c39c25$5c739eb0$7aa09044@NewDell>

This was why I liked rubys ability to return 2 values from a function. I
could return a true/false as to wether the function did its job, and if
it didn't I could return an array of strings detailing the error.

This works especially well in situations where the application under
test returns textual errors, which can then be sraped from the screen
and returned. For example, consider the testing of  'create user' 

The top level test calls the 'testCreateUser' function.

This attemps to create a duplicate username, the application under test
returns with a message such as 'This user already exists, please enter a
new user name'

The testCreateUser is able to grab this text from the application, and
return it to the higher level test. This can now examine the array of
strings returned, and will recognize this error, and act on it
accordingly

This is especially good for -ve testing, as we know we will fail, but we
can now check to see that we failed for the correct reason. For an
exception based system, we might be able to examine the exception to
discover this - maybe we throw a runtime exception and detail the error
in some additional text.

Another problem I came across with test::unit was its handling of
unexpected exceptions. This is probably due to my lack of experience
with it, but it seemed I had to wrap every assert in begin/rescue to
stop the program ending.

Ive been reaading up on ant today, and am tending towards the exception
based system, purely so that we could create an ant task for running gui
automation, and the exception based system more closely matches what is
currently being used.

Paul


-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Bret Pettichord
Sent: 26 October 2003 11:16
To: wtr-general at rubyforge.org
Subject: RE: [Wtr-general] Iecontroller with exceptions


At 12:07 PM 10/26/2003, Paul Rogers wrote:
>That is how Ive done it in the latest controller. One of the big 
>advantages of returning strings as the error rather than throwing 
>exceptions, was the ability to do negative testing with virtually no 
>extra over head. Using exceptions is probably a more correct way of 
>doing it but seems to me to generate a lot of extra overhead.

Have you tried using assert_raises?

I believe that signalling errors by return code requires more overhead 
because wrapper functions must be sure to check and return the values 
returned by the functions they call. Not doing this, or not doing this 
correctly, can cause testsuites to ignore triggered errors - a fatal
flaw.


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com

_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From chrismo at clabs.org  Sun Oct 26 23:20:11 2003
From: chrismo at clabs.org (Chris Morris)
Date: Sun, 26 Oct 2003 22:20:11 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <000001c39c25$5c739eb0$7aa09044@NewDell>
References: <000001c39c25$5c739eb0$7aa09044@NewDell>
Message-ID: <3F9C9CFB.2030503@clabs.org>

Paul Rogers wrote:

>Another problem I came across with test::unit was its handling of
>unexpected exceptions. This is probably due to my lack of experience
>with it, but it seemed I had to wrap every assert in begin/rescue to
>stop the program ending.
>  
>

This is intentional, as far as I know. An exception will stop the
current test from continuing, but other tests in the suite should still
run.

As I understand the larger approach -- every test should pass except
when something's broken. What's important about keeping the test running
in the instance when the test fails?

In all the suites I've built, by and large all tests remain successful,
so how I detect a failure (by boolean result or exception) just doesn't
seem to matter too much. When tests fail, I'm never concerned with
keeping that test running -- I go in, find out what broke, fix it and
get it passing again.

-- 
Chris
http://clabs.org/blogki






From paul.rogers at shaw.ca  Sun Oct 26 23:31:53 2003
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Sun, 26 Oct 2003 21:31:53 -0700
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <3F9C9CFB.2030503@clabs.org>
Message-ID: <000001c39c43$3fad26c0$7aa09044@NewDell>

So Im probably doing something wrong with test::unit - this makes me
feel a whole lot better.

As for keeping tests running, when testing an application (as opposed to
unit testing) , I want to find as many problems as I can in this build.
So its important the test continue.

Paul

-----Original Message-----
From: wtr-general-bounces at rubyforge.org
[mailto:wtr-general-bounces at rubyforge.org] On Behalf Of Chris Morris
Sent: 26 October 2003 21:20
To: wtr-general at rubyforge.org
Subject: Re: [Wtr-general] Iecontroller with exceptions


Paul Rogers wrote:

>Another problem I came across with test::unit was its handling of 
>unexpected exceptions. This is probably due to my lack of experience 
>with it, but it seemed I had to wrap every assert in begin/rescue to 
>stop the program ending.
>  
>

This is intentional, as far as I know. An exception will stop the
current test from continuing, but other tests in the suite should still
run.

As I understand the larger approach -- every test should pass except
when something's broken. What's important about keeping the test running
in the instance when the test fails?

In all the suites I've built, by and large all tests remain successful,
so how I detect a failure (by boolean result or exception) just doesn't
seem to matter too much. When tests fail, I'm never concerned with
keeping that test running -- I go in, find out what broke, fix it and
get it passing again.

-- 
Chris
http://clabs.org/blogki




_______________________________________________
Wtr-general mailing list
Wtr-general at rubyforge.org
http://rubyforge.org/mailman/listinfo/wtr-general



From bret at pettichord.com  Mon Oct 27 14:26:15 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Mon, 27 Oct 2003 13:26:15 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <000001c39c43$3fad26c0$7aa09044@NewDell>
References: <3F9C9CFB.2030503@clabs.org>
Message-ID: <5.1.0.14.2.20031027132343.02188b30@127.0.0.1>

At 10:31 PM 10/26/2003, Paul Rogers wrote:
>As for keeping tests running, when testing an application (as opposed to
>unit testing) , I want to find as many problems as I can in this build.
>So its important the test continue.

Well, you want the other tests to continue.

Once a test fails, all you want is a teardown so that you can run 
additional tests.

It may be that you are putting too many things into a test.

I've seen people put too much trouble into conditional logic to workaround 
potential errors. Is this what you are doing? Personally, i avoid this as 
much as possible because it's so easy to introduce silent errors.

Can you show us your tests?

Bret


_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Mon Oct 27 18:21:23 2003
From: chrismo at clabs.org (Chris Morris)
Date: Mon, 27 Oct 2003 17:21:23 -0600
Subject: [Wtr-general] Iecontroller with exceptions
In-Reply-To: <000001c39c25$5c739eb0$7aa09044@NewDell>
References: <000001c39c25$5c739eb0$7aa09044@NewDell>
Message-ID: <3F9DA873.9080709@clabs.org>

Paul Rogers wrote:

>This was why I liked rubys ability to return 2 values from a function. I
>could return a true/false as to wether the function did its job, and if
>it didn't I could return an array of strings detailing the error.
>  
>

An exception has all of this in one shot, and you don't have to
specifically handle it. Something goes wrong, it's raised, your code
won't continue on, and all the info you need (hopefully) is contained
within the exception object.

-- 
Chris
http://clabs.org/blogki






From paul.rogers at shaw.ca  Mon Oct 27 19:23:27 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Mon, 27 Oct 2003 17:23:27 -0700
Subject: [Wtr-general] Iecontroller with exceptions
Message-ID: <acddb9acec1f.acec1facddb9@shaw.ca>

Ok! Im Convinced! We do it as exceptions!

There are times that the other method is useful, so what I  will do is add the message array stuff as a class variable to the ie class. That way the stuff I am interested in is still available, and I keep everyone else happy too :-)

On a different note - Ive been battling with frames today. I think I have all the issues resolved, but is there a way in javascript ( as that is almost identical to what I would be doing in ruby) to dynamically get a parent frame name?

something like:

self.getParent

Thanks

Paul

----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Monday, October 27, 2003 4:21 pm
Subject: Re: [Wtr-general] Iecontroller with exceptions

> Paul Rogers wrote:
> 
> >This was why I liked rubys ability to return 2 values from a 
> function. I
> >could return a true/false as to wether the function did its job, 
> and if
> >it didn't I could return an array of strings detailing the error.
> >  
> >
> 
> An exception has all of this in one shot, and you don't have to
> specifically handle it. Something goes wrong, it's raised, your code
> won't continue on, and all the info you need (hopefully) is contained
> within the exception object.
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
> 



From groupmember at peterw.nl  Tue Oct 28 05:20:55 2003
From: groupmember at peterw.nl (groupmember at peterw.nl)
Date: Tue, 28 Oct 2003 11:20:55 +0100
Subject: [Wtr-general] IE playback facilities?
Message-ID: <031c01c39d3d$2c6861b0$9700000a@PWLaptop>

Hello y'all!

does anyone have experience in recording / playback user interaction in IE using the cliecontroller.rb? 
may be this can help me to resolve the problem mentioned earlier regarding "getting at the right form"

thanks!

regards,

Peter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/wtr-general/attachments/20031028/6a0ea939/attachment.html 

From bret at pettichord.com  Wed Oct 29 19:56:45 2003
From: bret at pettichord.com (Bret Pettichord)
Date: Wed, 29 Oct 2003 18:56:45 -0600
Subject: [Wtr-general] related item
Message-ID: <5.1.0.14.2.20031029185618.022858d0@127.0.0.1>

Here's something related that i haven't read yet...

http://www.cq2.nl/Testdriving%20Internet%20Explorer

_____________________________________
  Bret Pettichord, Software Tester
    Book - www.testinglessons.com
    Consulting - www.pettichord.com
    Blog - www.io.com/~wazmo/blog
    Hotlist - www.testinghotlist.com



From chrismo at clabs.org  Thu Oct 30 09:20:56 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 30 Oct 2003 08:20:56 -0600
Subject: [Wtr-general] related item
In-Reply-To: <5.1.0.14.2.20031029185618.022858d0@127.0.0.1>
References: <5.1.0.14.2.20031029185618.022858d0@127.0.0.1>
Message-ID: <3FA11E48.7030800@clabs.org>

Bret Pettichord wrote:

> Here's something related that i haven't read yet...
>
> http://www.cq2.nl/Testdriving%20Internet%20Explorer


Interesting, though I don't have time to check it out right now. Thanks 
for posting it.

-- 
Chris
http://clabs.org/blogki




From chrismo at clabs.org  Thu Oct 30 18:44:05 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 30 Oct 2003 17:44:05 -0600
Subject: [Wtr-general] IE playback facilities?
In-Reply-To: <031c01c39d3d$2c6861b0$9700000a@PWLaptop>
References: <031c01c39d3d$2c6861b0$9700000a@PWLaptop>
Message-ID: <3FA1A245.3080201@clabs.org>

groupmember at peterw.nl wrote:

> Hello y'all!
>  
> does anyone have experience in recording / playback user interaction 
> in IE using the cliecontroller.rb?
> may be this can help me to resolve the problem mentioned earlier 
> regarding "getting at the right form"

Sorry, there's nothing in the lib to do recording of actions.

Sorry also that I never replied to your earlier post. Lemme take a look 
at it again...

-- 
Chris
http://clabs.org/blogki





From chrismo at clabs.org  Thu Oct 30 18:47:54 2003
From: chrismo at clabs.org (Chris Morris)
Date: Thu, 30 Oct 2003 17:47:54 -0600
Subject: [Wtr-general] Getting at the right form
In-Reply-To: <00b301c39a38$f00ff6d0$2f02a8c0@PWLaptop>
References: <00b301c39a38$f00ff6d0$2f02a8c0@PWLaptop>
Message-ID: <3FA1A32A.8010201@clabs.org>

groupmember at peterw.nl wrote:

>  
>
> I have an application that I want to test with ruby and cliecontroller.rb 
> the thing is that the url <http://url> to test is always creating a 
> popup window and then closes the original IE window. 
> I would like to get to the form property of the newly created window 
> but I don't know how to get this done with the COM model 
>  
> anyone?


Ah, I'd thought I might need to look at the code -- but I hadn't really 
groked what you were trying to do.

I have no idea how to jump from window to window, even if the first 
window isn't shut.

You might look at the DOM wrapper code I've got in cliec -- there's a 
way you can dump the entire DOM tree to the file, see the readme in the 
dist (actually the readme may only be in the CVS). With the entire DOM 
dumped, maybe that'll give you a clue.

-- 
Chris
http://clabs.org/blogki





From paul.rogers at shaw.ca  Fri Oct 31 09:45:26 2003
From: paul.rogers at shaw.ca (paul.rogers at shaw.ca)
Date: Fri, 31 Oct 2003 07:45:26 -0700
Subject: [Wtr-general] Getting at the right form
Message-ID: <e5356de54372.e54372e5356d@shaw.ca>

I think there is a bug in  the ruby win32ole library that prevents us getting to new windows. I have started writing some code to demonstrate it and hopefully get it fixed. 

It seems that new windows dont appear in the frames collection of a document (although the IE docs seems to say it should)

----- Original Message -----
From: Chris Morris <chrismo at clabs.org>
Date: Thursday, October 30, 2003 4:47 pm
Subject: Re: [Wtr-general] Getting at the right form

> groupmember at peterw.nl wrote:
> 
> >  
> >
> > I have an application that I want to test with ruby and 
> cliecontroller.rb 
> > the thing is that the url <" target="l">http://url> to test is 
> always creating a 
> > popup window and then closes the original IE window. 
> > I would like to get to the form property of the newly created 
> window 
> > but I don't know how to get this done with the COM model 
> >  
> > anyone?
> 
> 
> Ah, I'd thought I might need to look at the code -- but I hadn't 
> really 
> groked what you were trying to do.
> 
> I have no idea how to jump from window to window, even if the 
> first 
> window isn't shut.
> 
> You might look at the DOM wrapper code I've got in cliec -- 
> there's a 
> way you can dump the entire DOM tree to the file, see the readme 
> in the 
> dist (actually the readme may only be in the CVS). With the entire 
> DOM 
> dumped, maybe that'll give you a clue.
> 
> -- 
> Chris
> http://clabs.org/blogki
> 
> 
> 
> _______________________________________________
> Wtr-general mailing list
> Wtr-general at rubyforge.org
> http://rubyforge.org/mailman/listinfo/wtr-general
>