From djberg96 at gmail.com  Fri Oct  6 16:02:38 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Fri, 06 Oct 2006 14:02:38 -0600
Subject: [Win32utils-devel] Need some help with latest win32-mmap
Message-ID: <4526B65E.6070508@gmail.com>

Hi all,

I've got the latest win32-mmap code checked into CVS.  Unfortunately, it 
seems that I'm not able to open an existing mapping and retrieve set 
data.  Below is a simple example that seems like it ought to work but 
doesn't.  Any ideas?

# map1.rb
require 'win32/mmap'
include Win32

mmap = MMap.new(:name => 'alpha', :size => 20)
p mmap.address

mmap.test = "hello"

mmap.close

# map2.rb
# mmap = MMap.open('alpha') # Boom!

mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far...
p mmap.test # Boom!

mmap.close

I tried using MMap.open and MMap.new.  Neither seems to work as 
expected.  What am I doing wrong?

Thanks,

Dan

PS - The code that actually gets and sets shared data is in method_missing.

From phasis at gmail.com  Mon Oct  9 01:02:51 2006
From: phasis at gmail.com (Heesob Park)
Date: Mon, 9 Oct 2006 14:02:51 +0900
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <4526B65E.6070508@gmail.com>
References: <4526B65E.6070508@gmail.com>
Message-ID: <aea95bc50610082202r8965ae2vcf884d39012e3c34@mail.gmail.com>

Hi,

2006/10/7, Daniel Berger <djberg96 at gmail.com>:
>
> Hi all,
>
> I've got the latest win32-mmap code checked into CVS.  Unfortunately, it
> seems that I'm not able to open an existing mapping and retrieve set
> data.  Below is a simple example that seems like it ought to work but
> doesn't.  Any ideas?
>
> # map1.rb
> require 'win32/mmap'
> include Win32
>
> mmap = MMap.new(:name => 'alpha', :size => 20)
> p mmap.address
>
> mmap.test = "hello"
>
> mmap.close
>
> # map2.rb
> # mmap = MMap.open('alpha') # Boom!
>
> mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far...
> p mmap.test # Boom!
>
> mmap.close
>
> I tried using MMap.open and MMap.new.  Neither seems to work as
> expected.  What am I doing wrong?


That's strange.
I downloaded and tested the win32-mmap and your map1.rb and map2.rb
It runs ok and seems to work as expected.

Regards,

Park Heesob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061009/19f9e144/attachment.html 

From Daniel.Berger at qwest.com  Mon Oct  9 16:47:44 2006
From: Daniel.Berger at qwest.com (Daniel Berger)
Date: Mon, 09 Oct 2006 14:47:44 -0600
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <aea95bc50610082202r8965ae2vcf884d39012e3c34@mail.gmail.com>
References: <4526B65E.6070508@gmail.com>
	<aea95bc50610082202r8965ae2vcf884d39012e3c34@mail.gmail.com>
Message-ID: <452AB570.2070105@qwest.com>

Heesob Park wrote:
> Hi,
> 
> 2006/10/7, Daniel Berger <djberg96 at gmail.com <mailto:djberg96 at gmail.com>>:
> 
>     Hi all,
> 
>     I've got the latest win32-mmap code checked into CVS.  Unfortunately, it
>     seems that I'm not able to open an existing mapping and retrieve set
>     data.  Below is a simple example that seems like it ought to work but
>     doesn't.  Any ideas?
> 
>     # map1.rb
>     require 'win32/mmap'
>     include Win32
> 
>     mmap = MMap.new(:name => 'alpha', :size => 20)
>     p mmap.address
> 
>     mmap.test = "hello"
> 
>     mmap.close
> 
>     # map2.rb
>     # mmap = MMap.open('alpha') # Boom!
> 
>     mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far...
>     p mmap.test # Boom!
> 
>     mmap.close
> 
>     I tried using MMap.open and MMap.new.  Neither seems to work as
>     expected.  What am I doing wrong?
> 
>  
> That's strange.
> I downloaded and tested the win32-mmap and your map1.rb and map2.rb
> It runs ok and seems to work as expected.
>  
> Regards,
>  
> Park Heesob

Really?  When you run map2.rb you see "hello"?

When I run map2.rb I get the same address.  But when I run map2.rb I get an 
error at mmap.rb, line 270: 'load - marshal data too short'.  It seems that I'm 
getting an empty buffer from the memcpy, but I don't understand why.

I just tried this at work, too - same result.

Regards,

Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From phasis at gmail.com  Mon Oct  9 22:02:21 2006
From: phasis at gmail.com (Heesob Park)
Date: Tue, 10 Oct 2006 11:02:21 +0900
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <452AB570.2070105@qwest.com>
References: <4526B65E.6070508@gmail.com>
	<aea95bc50610082202r8965ae2vcf884d39012e3c34@mail.gmail.com>
	<452AB570.2070105@qwest.com>
Message-ID: <aea95bc50610091902m6dec7443idb73cbe33f43eaec@mail.gmail.com>

Hi,

2006/10/10, Daniel Berger <Daniel.Berger at qwest.com>:

> Heesob Park wrote:
> > Hi,
> >
> > 2006/10/7, Daniel Berger <djberg96 at gmail.com <mailto:djberg96 at gmail.com
> >>:
> >
> >     Hi all,
> >
> >     I've got the latest win32-mmap code checked into
> CVS.  Unfortunately, it
> >     seems that I'm not able to open an existing mapping and retrieve set
> >     data.  Below is a simple example that seems like it ought to work
> but
> >     doesn't.  Any ideas?
> >
> >     # map1.rb
> >     require 'win32/mmap'
> >     include Win32
> >
> >     mmap = MMap.new(:name => 'alpha', :size => 20)
> >     p mmap.address
> >
> >     mmap.test = "hello"
> >
> >     mmap.close
> >
> >     # map2.rb
> >     # mmap = MMap.open('alpha') # Boom!
> >
> >     mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far...
> >     p mmap.test # Boom!
> >
> >     mmap.close
> >
> >     I tried using MMap.open and MMap.new.  Neither seems to work as
> >     expected.  What am I doing wrong?
> >
> >
> > That's strange.
> > I downloaded and tested the win32-mmap and your map1.rb and map2.rb
> > It runs ok and seems to work as expected.
> >
> > Regards,
> >
> > Park Heesob
>
> Really?  When you run map2.rb you see "hello"?
>
> When I run map2.rb I get the same address.  But when I run map2.rb I get
> an
> error at mmap.rb, line 270: 'load - marshal data too short'.  It seems
> that I'm
> getting an empty buffer from the memcpy, but I don't understand why.
>
> I just tried this at work, too - same result.


As a matter of fact I modified map1.rb like this:

require 'win32/mmap'
include Win32

mmap = MMap.new(:name => 'alpha', :size => 20)
p mmap.address

mmap.test = "hello"
sleep 1000
mmap.close

And run map1.rb in one console and run map2.rb in another console.
When running map2.rb only, I also get "marshal data too short" error.
Regards,

Park Heesob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061010/a0585682/attachment.html 

From johan.nilsson at esrange.ssc.se  Tue Oct 10 02:51:29 2006
From: johan.nilsson at esrange.ssc.se (Johan Nilsson)
Date: Tue, 10 Oct 2006 08:51:29 +0200
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <4526B65E.6070508@gmail.com>
Message-ID: <6D2C2D1D3D1E8040AEFA12C90600868201104AF3@rbe-mail1.ssc.internal>

Hi, 

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Daniel Berger
> Sent: den 6 oktober 2006 22:03
> To: win32utils-devel at rubyforge.org
> Subject: [Win32utils-devel] Need some help with latest win32-mmap
> 
> Hi all,
> 
> I've got the latest win32-mmap code checked into CVS.  
> Unfortunately, it seems that I'm not able to open an existing 
> mapping and retrieve set data.  Below is a simple example 
> that seems like it ought to work but doesn't.  Any ideas?
> 
> # map1.rb
> require 'win32/mmap'
> include Win32
> 
> mmap = MMap.new(:name => 'alpha', :size => 20) p mmap.address
> 
> mmap.test = "hello"
> 
> mmap.close
> 
> # map2.rb
> # mmap = MMap.open('alpha') # Boom!
> 
> mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far...
> p mmap.test # Boom!
> 
> mmap.close
> 
> I tried using MMap.open and MMap.new.  Neither seems to work 
> as expected.  What am I doing wrong?

I haven't checked the implementation and this is probably too obvious to
be the culprit, but:

If map1.rb runs to completion and closes the mmap before map2.rb can
attach by opening it, the memory map won't exist unless you are using
something else than the paging file as backing storage. The Win32 mmaps
are reference counted through their handles as pretty much everything
else.

Also, as you can see it that using MMap.new from map2.rb works as the
first step, but as this only creates a new identically named mmap there
is no 'test' data there. Hence the 'boom' I'd guess.

HTH,

Johan Nilsson


From Daniel.Berger at qwest.com  Tue Oct 10 09:43:18 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Tue, 10 Oct 2006 08:43:18 -0500
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <6D2C2D1D3D1E8040AEFA12C90600868201104AF3@rbe-mail1.ssc.internal>
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E871@QTOMAE2K3M01.AD.QINTRA.COM>

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Johan Nilsson
> Sent: Tuesday, October 10, 2006 12:51 AM
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
> 
> 
> Hi, 
> 
> > -----Original Message-----
> > From: win32utils-devel-bounces at rubyforge.org
> > [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> > Daniel Berger
> > Sent: den 6 oktober 2006 22:03
> > To: win32utils-devel at rubyforge.org
> > Subject: [Win32utils-devel] Need some help with latest win32-mmap
> > 
> > Hi all,
> > 
> > I've got the latest win32-mmap code checked into CVS.
> > Unfortunately, it seems that I'm not able to open an existing 
> > mapping and retrieve set data.  Below is a simple example 
> > that seems like it ought to work but doesn't.  Any ideas?
> > 
> > # map1.rb
> > require 'win32/mmap'
> > include Win32
> > 
> > mmap = MMap.new(:name => 'alpha', :size => 20) p mmap.address
> > 
> > mmap.test = "hello"
> > 
> > mmap.close
> > 
> > # map2.rb
> > # mmap = MMap.open('alpha') # Boom!
> > 
> > mmap = MMap.new(:name => 'alpha', :size => 20) # Ok so far... p 
> > mmap.test # Boom!
> > 
> > mmap.close
> > 
> > I tried using MMap.open and MMap.new.  Neither seems to work
> > as expected.  What am I doing wrong?
> 
> I haven't checked the implementation and this is probably too 
> obvious to be the culprit, but:
> 
> If map1.rb runs to completion and closes the mmap before 
> map2.rb can attach by opening it, the memory map won't exist 
> unless you are using something else than the paging file as 
> backing storage. The Win32 mmaps are reference counted 
> through their handles as pretty much everything else.
> 
> Also, as you can see it that using MMap.new from map2.rb 
> works as the first step, but as this only creates a new 
> identically named mmap there is no 'test' data there. Hence 
> the 'boom' I'd guess.
> 
> HTH,
> 
> Johan Nilsson

Yes, you and Heesob are both right.  I was thinking the value stayed
resident.  Once I stuck a sleep call in there and ran it from a second
terminal it worked fine.  Duh.

In other news, it looks like I need to set a default size for MMap.open.
What sounds reasonabl? 1k?

Thanks,

Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From johan.nilsson at esrange.ssc.se  Wed Oct 11 03:13:03 2006
From: johan.nilsson at esrange.ssc.se (Johan Nilsson)
Date: Wed, 11 Oct 2006 09:13:03 +0200
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <39AA6550E5AA554AB1456707D6E5563D0259E871@QTOMAE2K3M01.AD.QINTRA.COM>
Message-ID: <6D2C2D1D3D1E8040AEFA12C90600868201104B23@rbe-mail1.ssc.internal>

 

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Berger, Daniel
> Sent: den 10 oktober 2006 15:43
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
> 
> > -----Original Message-----
> > From: win32utils-devel-bounces at rubyforge.org
> > [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of Johan 
> > Nilsson
> > Sent: Tuesday, October 10, 2006 12:51 AM
> > To: Development and ideas for win32utils projects
> > Subject: Re: [Win32utils-devel] Need some help with latest 
> win32-mmap

[snip]

> > I haven't checked the implementation and this is probably 
> too obvious 
> > to be the culprit, but:
> > 
> > If map1.rb runs to completion and closes the mmap before 
> map2.rb can 
> > attach by opening it, the memory map won't exist unless you 
> are using 
> > something else than the paging file as backing storage. The Win32 
> > mmaps are reference counted through their handles as pretty much 
> > everything else.
> > 
> > Also, as you can see it that using MMap.new from map2.rb 
> works as the 
> > first step, but as this only creates a new identically named mmap 
> > there is no 'test' data there. Hence the 'boom' I'd guess.
> > 
> > HTH,
> > 
> > Johan Nilsson
> 
> Yes, you and Heesob are both right.  I was thinking the value 
> stayed resident.  Once I stuck a sleep call in there and ran 
> it from a second terminal it worked fine.  Duh.
> 
> In other news, it looks like I need to set a default size for 
> MMap.open.
> What sounds reasonabl? 1k?

I haven't tried this myself, and judging from the MSDN docs it's also
not entirely clear that it will work, but: Try using zero(0) as the size
when calling MapViewOfFile, and then use VirtualQuery to retrieve the
size of the view.

If you want to persist the values, use a file as the backing store
(could be an optional feature, e.g. :name => '/path/to/filename' ).

Oh, and use the "Global\" prefix for map names if you intend to share
the data between different terminal server sessions (Win2k and upwards).
But I guess you already do that internally.

Regards / Johan

From Daniel.Berger at qwest.com  Wed Oct 11 11:06:06 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Wed, 11 Oct 2006 10:06:06 -0500
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <6D2C2D1D3D1E8040AEFA12C90600868201104B23@rbe-mail1.ssc.internal>
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E877@QTOMAE2K3M01.AD.QINTRA.COM>

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Johan Nilsson
> Sent: Wednesday, October 11, 2006 1:13 AM
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap

<snip>

> > 
> > In other news, it looks like I need to set a default size for 
> > MMap.open.
> > What sounds reasonabl? 1k?
> 
> I haven't tried this myself, and judging from the MSDN docs it's also
> not entirely clear that it will work, but: Try using zero(0) 
> as the size
> when calling MapViewOfFile, and then use VirtualQuery to retrieve the
> size of the view.

Actually, I changed MMap.open to accept a hash of options that's simply
forwarded on to MMap.new (though the name is still a required argument),
so you can specify the size that way.  However, I think you're right
and, if it's possible, I'll do a VirtualQuery() to try to get the size
of the view when MMap.open is used if a size isn't specified explicitly.
That seems like a much friendlier way to go. :)

> If you want to persist the values, use a file as the backing store
> (could be an optional feature, e.g. :name => '/path/to/filename' ).

Well, there is a separate :file option.  I'm not sure I want to try to
guess whether :name refers to a file or not.  Or have I misunderstood?
 
> Oh, and use the "Global\" prefix for map names if you intend to share
> the data between different terminal server sessions (Win2k 
> and upwards).
> But I guess you already do that internally.

It will be up to the programmer to know details like that for the :name
option, though I can copy/paste the relevant section out of the MSDN
docs.

Thanks for the feedback.

- Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From Daniel.Berger at qwest.com  Wed Oct 11 14:39:45 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Wed, 11 Oct 2006 13:39:45 -0500
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <6D2C2D1D3D1E8040AEFA12C90600868201104B23@rbe-mail1.ssc.internal>
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E87C@QTOMAE2K3M01.AD.QINTRA.COM>

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Johan Nilsson
> Sent: Wednesday, October 11, 2006 1:13 AM
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap

<snip>
 
> I haven't tried this myself, and judging from the MSDN docs it's also
> not entirely clear that it will work, but: Try using zero(0) 
> as the size
> when calling MapViewOfFile, and then use VirtualQuery to retrieve the
> size of the view.

I've been tinkering with this a bit.  I think I'm doing something wrong.
Here's a private method I created to get the view size.  Note that this
requires modifying the "VirtualQuery" (in windows/memory.rb) method
signature to "LPL":

def get_view_size
   mbi = [0,0,0,0,0,0,0].pack('LLLLLLL') # MEMORY_BASIC_INFORMATION
   VirtualQuery(@base_address, mbi, mbi.size)         
   return mbi[12,4].unpack('L').first # RegionSize
end

Then, in MMap.new, I have this:

if @open
   size = get_view_size()
   @address = MapViewOfFileEx(@mh, @access, 0, 0, size, @base_address)
else
   @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
end

The first issue, I think, is that get_view_size always returns 65536
(64k).  I'm guessing 64k is a default amount of memory to allocate or
something.  Using the examples I provided previously, I would expect
this to be 20.

The second issue is that MapViewOfFileEx fails if I try to pass a size.
I think I'm misreading the docs.

I'd make some more commits so you could just check this out from
RubyForge, but I'm getting timeouts at the moment from rubyforge.org...

Regards,

Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From phasis at gmail.com  Wed Oct 11 21:09:05 2006
From: phasis at gmail.com (Heesob Park)
Date: Thu, 12 Oct 2006 10:09:05 +0900
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <39AA6550E5AA554AB1456707D6E5563D0259E87C@QTOMAE2K3M01.AD.QINTRA.COM>
References: <6D2C2D1D3D1E8040AEFA12C90600868201104B23@rbe-mail1.ssc.internal>
	<39AA6550E5AA554AB1456707D6E5563D0259E87C@QTOMAE2K3M01.AD.QINTRA.COM>
Message-ID: <aea95bc50610111809m5e75e373x71ecd43ffc1964f0@mail.gmail.com>

Hi,

2006/10/12, Berger, Daniel <Daniel.Berger at qwest.com>:
>
> > -----Original Message-----
> > From: win32utils-devel-bounces at rubyforge.org
> > [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of
> > Johan Nilsson
> > Sent: Wednesday, October 11, 2006 1:13 AM
> > To: Development and ideas for win32utils projects
> > Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
>
> <snip>
>
> > I haven't tried this myself, and judging from the MSDN docs it's also
> > not entirely clear that it will work, but: Try using zero(0)
> > as the size
> > when calling MapViewOfFile, and then use VirtualQuery to retrieve the
> > size of the view.
>
> I've been tinkering with this a bit.  I think I'm doing something wrong.
> Here's a private method I created to get the view size.  Note that this
> requires modifying the "VirtualQuery" (in windows/memory.rb) method
> signature to "LPL":
>
> def get_view_size
>   mbi = [0,0,0,0,0,0,0].pack('LLLLLLL') # MEMORY_BASIC_INFORMATION
>   VirtualQuery(@base_address, mbi, mbi.size)
>   return mbi[12,4].unpack('L').first # RegionSize
> end


should be:

def get_view_size
  mbi = [0,0,0,0,0,0,0].pack('LLLLLLL') # MEMORY_BASIC_INFORMATION
  VirtualQuery(@address, mbi, mbi.size)
  return mbi[12,4].unpack('L').first # RegionSize
end


Then, in MMap.new, I have this:
>
> if @open
>   size = get_view_size()
>   @address = MapViewOfFileEx(@mh, @access, 0, 0, size, @base_address)
> else
>   @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
> end


should be:

if @open
  @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
  @size = get_view_size()
else
  @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
end


The first issue, I think, is that get_view_size always returns 65536
> (64k).  I'm guessing 64k is a default amount of memory to allocate or
> something.  Using the examples I provided previously, I would expect
> this to be 20.
>
> The second issue is that MapViewOfFileEx fails if I try to pass a size.
> I think I'm misreading the docs.
>
> I'd make some more commits so you could just check this out from
> RubyForge, but I'm getting timeouts at the moment from rubyforge.org...
>
> Regards,
>
> Dan


It seems the memory map in windows aligns 4k boundary.
When created with 20, get_view_size returns 4096
When created with 20000, get_view_size returns 20480.

Regards,

Park Heesob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061012/ce212b83/attachment.html 

From djberg96 at gmail.com  Wed Oct 11 23:18:36 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Wed, 11 Oct 2006 21:18:36 -0600
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <aea95bc50610111809m5e75e373x71ecd43ffc1964f0@mail.gmail.com>
References: <6D2C2D1D3D1E8040AEFA12C90600868201104B23@rbe-mail1.ssc.internal>
	<39AA6550E5AA554AB1456707D6E5563D0259E87C@QTOMAE2K3M01.AD.QINTRA.COM>
	<aea95bc50610111809m5e75e373x71ecd43ffc1964f0@mail.gmail.com>
Message-ID: <6037b70c0610112018j276a40b9t41759040f8dc9c93@mail.gmail.com>

On 10/11/06, Heesob Park <phasis at gmail.com> wrote:
>
> Hi,
>
> 2006/10/12, Berger, Daniel <Daniel.Berger at qwest.com>:
> >
> > > -----Original Message-----
> > > From: win32utils-devel-bounces at rubyforge.org
> > > [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of
> > > Johan Nilsson
> > > Sent: Wednesday, October 11, 2006 1:13 AM
> > > To: Development and ideas for win32utils projects
> > > Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
> >
> > <snip>
> >
> > > I haven't tried this myself, and judging from the MSDN docs it's also
> > > not entirely clear that it will work, but: Try using zero(0)
> > > as the size
> > > when calling MapViewOfFile, and then use VirtualQuery to retrieve the
> > > size of the view.
> >
> > I've been tinkering with this a bit.  I think I'm doing something wrong.
> > Here's a private method I created to get the view size.  Note that this
> > requires modifying the "VirtualQuery" (in windows/memory.rb) method
> > signature to "LPL":
> >
> > def get_view_size
> >   mbi = [0,0,0,0,0,0,0].pack('LLLLLLL') # MEMORY_BASIC_INFORMATION
> >   VirtualQuery(@base_address, mbi, mbi.size)
> >   return mbi[12,4].unpack('L').first # RegionSize
> > end
>
>
> should be:
>
> def get_view_size
>   mbi = [0,0,0,0,0,0,0].pack('LLLLLLL') # MEMORY_BASIC_INFORMATION
>   VirtualQuery(@address, mbi, mbi.size)
>   return mbi[12,4].unpack('L').first # RegionSize
> end
>
>
> Then, in MMap.new, I have this:
> >
> > if @open
> >   size = get_view_size()
> >   @address = MapViewOfFileEx(@mh, @access, 0, 0, size, @base_address)
> > else
> >   @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
> > end
>
>
> should be:
>
> if @open
>   @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
>   @size = get_view_size()
> else
>   @address = MapViewOfFileEx(@mh, @access, 0, 0, 0, @base_address)
> end
>
>
> The first issue, I think, is that get_view_size always returns 65536
> > (64k).  I'm guessing 64k is a default amount of memory to allocate or
> > something.  Using the examples I provided previously, I would expect
> > this to be 20.
> >
> > The second issue is that MapViewOfFileEx fails if I try to pass a size.
> > I think I'm misreading the docs.
> >
> > I'd make some more commits so you could just check this out from
> > RubyForge, but I'm getting timeouts at the moment from rubyforge.org...
> >
> > Regards,
> >
> > Dan
>
>
> It seems the memory map in windows aligns 4k boundary.
> When created with 20, get_view_size returns 4096
> When created with 20000, get_view_size returns 20480.
>
> Regards,
>
> Park Heesob
>

Many thanks.  I have checked the changes into CVS, along with the semaphore
locking code.

All, please take a look and let me know if you think I'm missing anything or
if there are any other interface changes you think need to be made.  I
wasn't sure about the method_missing approach, but I think it's nicer than
the old getvar/setvar approach.  I can always add that back in if necessary.

If all looks good, I'll try to get a release out this weekend.

Regards,

Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061011/9dddc674/attachment.html 

From johan.nilsson at esrange.ssc.se  Thu Oct 12 02:45:35 2006
From: johan.nilsson at esrange.ssc.se (Johan Nilsson)
Date: Thu, 12 Oct 2006 08:45:35 +0200
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <aea95bc50610111809m5e75e373x71ecd43ffc1964f0@mail.gmail.com>
Message-ID: <6D2C2D1D3D1E8040AEFA12C90600868201104B4E@rbe-mail1.ssc.internal>


>	From: win32utils-devel-bounces at rubyforge.org
[mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of Heesob Park
>	Sent: den 12 oktober 2006 03:09
>	To: Development and ideas for win32utils projects
>	Subject: Re: [Win32utils-devel] Need some help with latest
win32-mmap
>	

[snip]	

	 
>	It seems the memory map in windows aligns 4k boundary.
>	When created with 20, get_view_size returns 4096
>	When created with 20000, get_view_size returns 20480. 

Actually, it aligns to the nearest page boundary (which is not
necessarily 4k). Use GetSystemInformation to retrieve SYSTEM_INFO
(dwPageSize member).

/ Johan

From johan.nilsson at esrange.ssc.se  Thu Oct 12 02:48:14 2006
From: johan.nilsson at esrange.ssc.se (Johan Nilsson)
Date: Thu, 12 Oct 2006 08:48:14 +0200
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <39AA6550E5AA554AB1456707D6E5563D0259E877@QTOMAE2K3M01.AD.QINTRA.COM>
Message-ID: <6D2C2D1D3D1E8040AEFA12C90600868201104B50@rbe-mail1.ssc.internal>

 

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Berger, Daniel
> Sent: den 11 oktober 2006 17:06
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
> 
> > -----Original Message-----
> > From: win32utils-devel-bounces at rubyforge.org
> > [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of Johan 
> > Nilsson
> > Sent: Wednesday, October 11, 2006 1:13 AM
> > To: Development and ideas for win32utils projects
> > Subject: Re: [Win32utils-devel] Need some help with latest 
> win32-mmap

[snip]

> 
> > If you want to persist the values, use a file as the backing store 
> > (could be an optional feature, e.g. :name => '/path/to/filename' ).
> 
> Well, there is a separate :file option.  I'm not sure I want 
> to try to guess whether :name refers to a file or not.  Or 
> have I misunderstood?

I'm sorry - I only replied using the information in previous mails.
Haven't actually tried to use mmap or check out the examples.

/ Johan

From Daniel.Berger at qwest.com  Thu Oct 12 09:41:11 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Thu, 12 Oct 2006 08:41:11 -0500
Subject: [Win32utils-devel] Need some help with latest win32-mmap
In-Reply-To: <6D2C2D1D3D1E8040AEFA12C90600868201104B4E@rbe-mail1.ssc.internal>
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E87F@QTOMAE2K3M01.AD.QINTRA.COM>

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Johan Nilsson
> Sent: Thursday, October 12, 2006 12:46 AM
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] Need some help with latest win32-mmap
> 
> 
> 
> >	From: win32utils-devel-bounces at rubyforge.org
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Heesob Park
> >	Sent: den 12 oktober 2006 03:09
> >	To: Development and ideas for win32utils projects
> >	Subject: Re: [Win32utils-devel] Need some help with latest
> win32-mmap
> >	
> 
> [snip]	
> 
> 	 
> >	It seems the memory map in windows aligns 4k boundary.
> >	When created with 20, get_view_size returns 4096
> >	When created with 20000, get_view_size returns 20480.
> 
> Actually, it aligns to the nearest page boundary (which is 
> not necessarily 4k). Use GetSystemInformation to retrieve 
> SYSTEM_INFO (dwPageSize member).

Right, I'll note that in the docs, thanks.

- Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From Daniel.Berger at qwest.com  Fri Oct 13 15:00:09 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Fri, 13 Oct 2006 14:00:09 -0500
Subject: [Win32utils-devel] win32-mmap - trying to marshal self
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E88E@QTOMAE2K3M01.AD.QINTRA.COM>

Hi all,

I realized the current implmentation has a problem - you can only get the last value set?  I realized, after looking at the old C code, that it actually stores values in a hash and marshals the hash, not the values themselves.

That seemed clunky to me, though.  I thought it would be more interesting if we just marshalled the entire mmap object and passed that back and forth.  Unfortunately, it doesn't seem to work.

For the getter in method_missing I simply did:

   marshal = Marshal.dump(self)

For the getter in method_missing I did:

   obj = memcpy(buf, @address, @size)
   val = object.send(obj.to_sym)
   instance_variable_set("@#{method}", Marshal.load(val))

But when I run the map1.rb and map2.rb scripts again, I get errors (and beeping!) with odd characters no less:

C:/Documents and Settings/djberge/workspace/win32-mmap/lib/win32/mmap.rb:334:in `send': undefined method o:?Win32::MMap?
@inheriti' for nil:NilClass (NoMethodError)
        from C:/Documents and Settings/djberge/workspace/win32-mmap/lib/win32/mmap.rb:334:in `method_missing'
        from map2.rb:8

Is this because of the Semaphore?  If we can't marshal self, that's fine, I'll just resort to the hash strategy, but I thought I would ask.

Thanks,

Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From djberg96 at gmail.com  Fri Oct 13 21:03:22 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Fri, 13 Oct 2006 19:03:22 -0600
Subject: [Win32utils-devel] win32-mmap - trying to marshal self
In-Reply-To: <39AA6550E5AA554AB1456707D6E5563D0259E88E@QTOMAE2K3M01.AD.QINTRA.COM>
References: <39AA6550E5AA554AB1456707D6E5563D0259E88E@QTOMAE2K3M01.AD.QINTRA.COM>
Message-ID: <6037b70c0610131803h28cadb2by8fe34da1e2a43b1f@mail.gmail.com>

On 10/13/06, Berger, Daniel <Daniel.Berger at qwest.com> wrote:
>
> Hi all,
>
> I realized the current implmentation has a problem - you can only get the
> last value set?  I realized, after looking at the old C code, that it
> actually stores values in a hash and marshals the hash, not the values
> themselves.
>
> That seemed clunky to me, though.  I thought it would be more interesting
> if we just marshalled the entire mmap object and passed that back and
> forth.  Unfortunately, it doesn't seem to work.


<snip>

Ah, nevermind.  I'll just pass a hash back and forth for now.  I want to get
something released this weekend so that folks get a chance to play with it
in time for RubyConf.

BTW, anyone on the list going to be at RubyConf?

- Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061013/a114414f/attachment.html 

From waynev at gmail.com  Sat Oct 14 02:05:42 2006
From: waynev at gmail.com (Wayne Vucenic)
Date: Fri, 13 Oct 2006 23:05:42 -0700
Subject: [Win32utils-devel] win32-mmap - trying to marshal self
In-Reply-To: <6037b70c0610131803h28cadb2by8fe34da1e2a43b1f@mail.gmail.com>
References: <39AA6550E5AA554AB1456707D6E5563D0259E88E@QTOMAE2K3M01.AD.QINTRA.COM>
	<6037b70c0610131803h28cadb2by8fe34da1e2a43b1f@mail.gmail.com>
Message-ID: <88c9ce410610132305q747590b9p149a91c831f571d@mail.gmail.com>

Hi Dan,

> BTW, anyone on the list going to be at RubyConf?

I enjoyed meeting you at RubyConf last year.  See you in Denver!

Wayne

On 10/13/06, Daniel Berger <djberg96 at gmail.com> wrote:
>
>
> On 10/13/06, Berger, Daniel <Daniel.Berger at qwest.com> wrote:
> > Hi all,
> >
> > I realized the current implmentation has a problem - you can only get the
> last value set?  I realized, after looking at the old C code, that it
> actually stores values in a hash and marshals the hash, not the values
> themselves.
> >
> > That seemed clunky to me, though.  I thought it would be more interesting
> if we just marshalled the entire mmap object and passed that back and forth.
>  Unfortunately, it doesn't seem to work.
>
> <snip>
>
> Ah, nevermind.  I'll just pass a hash back and forth for now.  I want to get
> something released this weekend so that folks get a chance to play with it
> in time for RubyConf.
>
> BTW, anyone on the list going to be at RubyConf?
>
> - Dan
>
>
>
> _______________________________________________
> win32utils-devel mailing list
> win32utils-devel at rubyforge.org
> http://rubyforge.org/mailman/listinfo/win32utils-devel
>
>

From noreply at rubyforge.org  Fri Oct 13 23:02:53 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Fri, 13 Oct 2006 23:02:53 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Bugs-5287 ] v 0.4.1 -
	Clipboard.set_data causes Segmentation fault
Message-ID: <20061014030255.1CE2F5241905@rubyforge.org>

Bugs item #5287, was opened at 2006-08-03 08:34
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=411&aid=5287&group_id=85

Category: win32-clipboard
Group: Code
>Status: Closed
>Resolution: Rejected
Priority: 3
Submitted By: Nobody (None)
Assigned to: Nobody (None)
Summary:  v 0.4.1 - Clipboard.set_data causes Segmentation fault

Initial Comment:
Whenever Clipboard.set_data is called a Segmentation fault occurs and ruby reports abnormal program shutdown.

You can discover this simply by running the test case.
I have observed this on two diffent systems.
Version 0.4.0 still works fine.

win32-clipboard-0.4.1
ruby 1.8.4
WinXP sp2

Edwin King

----------------------------------------------------------------------

>Comment By: Daniel Berger (djberg96)
Date: 2006-10-13 20:02

Message:
Rejected because I am almost certain the OP was using an old version of windows-pr and I can't reproduce it.

----------------------------------------------------------------------

Comment By: Daniel Berger (djberg96)
Date: 2006-08-04 17:28

Message:
Are you sure?

>ruby test\tc_clipboard.rb
Loaded suite test/tc_clipboard
Started
.............
Finished in 0.016 seconds.

13 tests, 36 assertions, 0 failures, 0 errors

What version of windows-pr are you using?

- Dan

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=411&aid=5287&group_id=85

From djberg96 at yahoo.com  Mon Oct 16 22:07:12 2006
From: djberg96 at yahoo.com (Daniel Berger)
Date: Mon, 16 Oct 2006 19:07:12 -0700 (PDT)
Subject: [Win32utils-devel] Fw: mmap 0.2.0 writing value from a child process
Message-ID: <20061017020712.10756.qmail@web50307.mail.yahoo.com>

I thought I would forward this to see if anyone has any suggestions (other than, "You shouldn't use this code in conjunction with fork").

Thanks,

Dan

----- Forwarded Message ----
From: "Park, Chong SuhX" <chong.suhx.park at intel.com>
To: Daniel Berger <djberg96 at yahoo.com>
Sent: Monday, October 16, 2006 3:59:55 PM
Subject: mmap 0.2.0 writing value from a child process




 
 


<!--

_filtered {font-family:Wingdings;
panose-1:5 0 0 0 0 0 0 0 0 0;}
_filtered {font-family:Batang;
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {font-family:"Trebuchet MS";
panose-1:2 11 6 3 2 2 2 2 2 4;}
_filtered {
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {
margin:1.0in 1.25in 1.0in 1.25in;}
_filtered {

}
_filtered {




text-indent:-.25in;
font-family:Wingdings;

}
_filtered {

}
_filtered {




text-indent:-.25in;
font-family:Wingdings;

}
_filtered {font-family:Wingdings;
panose-1:5 0 0 0 0 0 0 0 0 0;}
_filtered {font-family:Batang;
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
_filtered {font-family:"Trebuchet MS";
panose-1:2 11 6 3 2 2 2 2 2 4;}
_filtered {
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {
margin:1.0in 1.25in 1.0in 1.25in;}
_filtered {font-family:Wingdings;
panose-1:5 0 0 0 0 0 0 0 0 0;}
_filtered {font-family:Batang;
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
_filtered {font-family:"Trebuchet MS";
panose-1:2 11 6 3 2 2 2 2 2 4;}
_filtered {
panose-1:2 3 6 0 0 1 1 1 1 1;}
_filtered {
margin:1.0in 1.25in 1.0in 1.25in;}

 _filtered {font-family:Wingdings;
panose-1:5 0 0 0 0 0 0 0 0 0;}
 _filtered {font-family:Batang;
panose-1:2 3 6 0 0 1 1 1 1 1;}
 _filtered {font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
 _filtered {font-family:"Trebuchet MS";
panose-1:2 11 6 3 2 2 2 2 2 4;}
 _filtered {
panose-1:2 3 6 0 0 1 1 1 1 1;}
/* Style Definitions */
 p.MsoNormal, li.MsoNormal, div.MsoNormal
	{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman";}
a:link, span.MsoHyperlink
	{color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
	{color:purple;
text-decoration:underline;}
span.emailstyle17
	{font-family:"Trebuchet MS";
color:windowtext;
font-weight:normal;
font-style:normal;
text-decoration:none none;}
span.emailstyle18
	{font-family:"Trebuchet MS";
color:blue;
font-weight:normal;
font-style:normal;
text-decoration:none none;}
span.emailstyle19
	{font-family:"Trebuchet MS";
color:blue;
font-weight:normal;
font-style:normal;
text-decoration:none none;}
span.EmailStyle20
	{
font-family:"Trebuchet MS";
color:blue;
font-weight:normal;
font-style:normal;
text-decoration:none none;}
 _filtered {
margin:1.0in 1.25in 1.0in 1.25in;}
div.Section1
	{}
-->






Hi Dan,
 

  
 

First of all, I
sincerely appreciate your great contribution. Mmap 0.2.0 is critically
important piece of software for my project and would like to share with an
issue I am found today.  
 

  
 

I think it might
be a bug. Try following code and see if it is working. And then, remove comment
#  mmap.bar = 'My World' and run it again. You may find something interesting
 

  
 

pid1 = fork do
 

  
 

    mmap =
MMap.new(:name => 'alpha', :size => 2000)
 

    
 

    mmap.foo =
'Hello'
 

    mmap.bar =
'Your World'
 

  
 

    sleep 5
 

    mmap.close
 

  
 

end
 

  
 

sleep 1
 

  
 

pid2 = fork do
 

  
 

    mmap =
MMap.open('alpha')
 

  
 

  #  mmap.bar =
'My World'
 

  
 

    p mmap.foo
 

    p mmap.bar
 

  
 

    mmap.close
 

    
 

end
 







-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061016/c840a720/attachment-0001.html 

From djberg96 at yahoo.com  Thu Oct 19 19:42:47 2006
From: djberg96 at yahoo.com (Daniel Berger)
Date: Thu, 19 Oct 2006 16:42:47 -0700 (PDT)
Subject: [Win32utils-devel] Fw: Win32 OVERLAPPING IO
Message-ID: <20061019234247.57555.qmail@web50305.mail.yahoo.com>

Suggestions?

----- Forwarded Message ----
From: James Tucker <jt at ra66i.co.uk>
To: djberg96 at yahoo.com
Sent: Thursday, October 19, 2006 7:28:30 AM
Subject: Win32 OVERLAPPING IO

Hi Daniel,

    My name is James Tucker, I've been hacking in Ruby for only a few 
months now, however am an experienced general programmer in a wide 
variety of languages. My current focus is on a new development project 
involving an RS232 data protocol handler, which up until now I have 
programmed in ruby.

    We have a small problem with the ruby platform for this issue however, 
and it is related to several factors:

  - Ruby threads are green threads, IO blocks block all threads.
  - RS232 Comms on Windows using the Windows::File#open don't provide 
non overlapping I/O.
  - There is little easy to introduce documentation on how to use 
Windows::File#ReadFileEx in ruby.

    Ruby does however provide us with some major advantages. I programmed 
the stateful parser and threading system in under a day, and to date it 
serves perfectly as a one-way reader. Today however, we need to change 
this to continue this project.

    Besides my own testing code, I have been reading as much as possible 
around the topic, and one such thing which has really raised an eyebrow 
is this: http://msdn.microsoft.com/msdnmag/issues/02/10/NETSerialComm/

    In particular the following statement:
"Out-of-the-box, the only way of coding RS232 serial communications 
applications in the .NET environment is to import the outdated and 
somewhat limited MSComm ActiveX control. This article describes the 
development of a lean, multithreaded, and modern RS232 base class 
library in C# managed code. The library uses Platform Invocation 
Services to interact with the Win32 API directly. Application 
programmers can use the library from any .NET language through 
inheritance; the article explores examples written in C# and Visual 
Basic .NET."

    This would suggest that the only clean way to access RS232 non-blocking 
I/O in Windows is through this COM/Win32 API. I'm interested to know if:

  - The non-overlapping mode of Windows::File#ReadFileEx can achieve the 
same goal (if so, I would love to have some guidance on this, and in 
return I'll produce some clean documentation for you to add to the src).
  - If not, is there a way to hook the Win32 API using your code, in 
order to achieve the same goal?

    One final question, which I could not see a clear answer to reading on 
rubyforge, does upgrading the windows-pr gem upgrade all of the Windows 
namespace? I had some problems with Daemon a while back and am wondering 
if a fresh upgrade of windows-pr will solve this issue?

Many thanks for any time you can spare me,

James Tucker.

P.S. If you are able to get onto freenode, I'm 'raggi'.




-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3573 bytes
Desc: S/MIME Cryptographic Signature
Url : http://rubyforge.org/pipermail/win32utils-devel/attachments/20061019/dfe2fa07/attachment.bin 

From paul.rogers at shaw.ca  Thu Oct 19 22:27:55 2006
From: paul.rogers at shaw.ca (Paul Rogers)
Date: Thu, 19 Oct 2006 20:27:55 -0600
Subject: [Win32utils-devel] Fw: Win32 OVERLAPPING IO
References: <20061019234247.57555.qmail@web50305.mail.yahoo.com>
Message-ID: <019a01c6f3ef$79dabef0$6400a8c0@laptop>

some years ago I wrote a modbus simulator using ruby. Modbus is a protocol 
used on industrial measuring devices, like strain guages, thermomenters, 
flow meters ertc.

I cant rememebr exactly what I did or which libraries I used, but I had one 
ruby process that dealt with the rs232 comms which communicated with the 
main app using drb.

Paul

----- Original Message ----- 
From: "Daniel Berger" <djberg96 at yahoo.com>
To: <win32utils-devel at rubyforge.org>
Sent: Thursday, October 19, 2006 5:42 PM
Subject: [Win32utils-devel] Fw: Win32 OVERLAPPING IO


Suggestions?

----- Forwarded Message ----
From: James Tucker <jt at ra66i.co.uk>
To: djberg96 at yahoo.com
Sent: Thursday, October 19, 2006 7:28:30 AM
Subject: Win32 OVERLAPPING IO

Hi Daniel,

    My name is James Tucker, I've been hacking in Ruby for only a few
months now, however am an experienced general programmer in a wide
variety of languages. My current focus is on a new development project
involving an RS232 data protocol handler, which up until now I have
programmed in ruby.

    We have a small problem with the ruby platform for this issue however,
and it is related to several factors:

  - Ruby threads are green threads, IO blocks block all threads.
  - RS232 Comms on Windows using the Windows::File#open don't provide
non overlapping I/O.
  - There is little easy to introduce documentation on how to use
Windows::File#ReadFileEx in ruby.

    Ruby does however provide us with some major advantages. I programmed
the stateful parser and threading system in under a day, and to date it
serves perfectly as a one-way reader. Today however, we need to change
this to continue this project.

    Besides my own testing code, I have been reading as much as possible
around the topic, and one such thing which has really raised an eyebrow
is this: http://msdn.microsoft.com/msdnmag/issues/02/10/NETSerialComm/

    In particular the following statement:
"Out-of-the-box, the only way of coding RS232 serial communications
applications in the .NET environment is to import the outdated and
somewhat limited MSComm ActiveX control. This article describes the
development of a lean, multithreaded, and modern RS232 base class
library in C# managed code. The library uses Platform Invocation
Services to interact with the Win32 API directly. Application
programmers can use the library from any .NET language through
inheritance; the article explores examples written in C# and Visual
Basic .NET."

    This would suggest that the only clean way to access RS232 non-blocking
I/O in Windows is through this COM/Win32 API. I'm interested to know if:

  - The non-overlapping mode of Windows::File#ReadFileEx can achieve the
same goal (if so, I would love to have some guidance on this, and in
return I'll produce some clean documentation for you to add to the src).
  - If not, is there a way to hook the Win32 API using your code, in
order to achieve the same goal?

    One final question, which I could not see a clear answer to reading on
rubyforge, does upgrading the windows-pr gem upgrade all of the Windows
namespace? I had some problems with Daemon a while back and am wondering
if a fresh upgrade of windows-pr will solve this issue?

Many thanks for any time you can spare me,

James Tucker.

P.S. If you are able to get onto freenode, I'm 'raggi'.







--------------------------------------------------------------------------------


> _______________________________________________
> win32utils-devel mailing list
> win32utils-devel at rubyforge.org
> http://rubyforge.org/mailman/listinfo/win32utils-devel 



From djberg96 at gmail.com  Sun Oct 22 21:34:36 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Sun, 22 Oct 2006 19:34:36 -0600
Subject: [Win32utils-devel] Need some help with MMap#[] and MMap#[]=
Message-ID: <453C1C2C.8060900@gmail.com>

Hi all,

At RubyConf this year I talked about win32-mmap with Patrick Hurley and 
Ara Howard.  Ara mentioned that Guy's version of mmap (for Unix) 
contains some string-like methods that let you access pieces of the view 
like a string.  Patrick and I came up with this (which requires the 
latest windows-pr in cvs so that we can use a more flexible version of 
memcpy, as well as some constants for VirtualAlloc):

def [](offset, length)
   buf = 0.chr * length
   memcpy(buf, @address + offset, length)
   buf
end
      
def []=(offset, length, string)
   memcpy(@address + offset, string, string.length)
end

This works as is but Ara also mentioned that some methods, such as 
MMap#[]=, will autoexpand the allocation size if the offset + length is 
greater than the initial allocation size.  With the above implementation 
the string gets chopped:

mmap = MMap.new(:file => 'test.txt', :size => 3)
mmap[0,6] = "foobar" # only 'foo' gets written
mmap.close

I thought this was doable with VirtualAlloc, but I can't seem to get it 
to work:

def []=(offset, length, string)
   if offset + length > @size
      diff = offset + length - size
      addr = [@address + diff + 1].pack('L')
      if VirtualAlloc(addr, diff, MEM_COMMIT|MEM_RESERVE, 
PAGE_READWRITE) == 0
         raise Error, get_last_error
      end
   end
   memcpy(@address + offset, string, string.length)
end

When I try this, however, I get:

C:/ruby/lib/ruby/site_ruby/1.8/windows/memory.rb:82:in `call': NULL 
pointer given (ArgumentError)
        from C:/ruby/lib/ruby/site_ruby/1.8/windows/memory.rb:82:in 
`VirtualAlloc'
        from C:/Documents and 
Settings/djberge/workspace/win32-mmap/lib/win32/mmap.rb:309:in `[]='
        from test.rb:8

Am I using the right function? Do I have it declared correctly in 
windows-pr?  BTW, I also tried changed the function prototype for 
VirtualAlloc from 'PLLL' to just 'LLLL' and using the numeric address 
directly, but that didn't seem to help.

Any ideas?

Thanks,

Dan

PS - It was great to see the guys from win32utils-devel who showed up at 
RubyConf!

From djberg96 at gmail.com  Wed Oct 25 23:43:49 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Wed, 25 Oct 2006 21:43:49 -0600
Subject: [Win32utils-devel] The win32-service patch is committed - and now,
	some questions
Message-ID: <45402EF5.6060105@gmail.com>

Hi all,

I *finally* committed the patch from Patrick Hurley (you know, the one 
he submitted way back in June).  Looks like it works.  I've also 
revamped the code under the examples directory.  There's now a 
tdaemon.rb (the test daemon) and tdaemon_ctl.rb (the test daemon 
controller).

Before I release this into the wild I want to make sure I document how 
to use it properly.  Based on what I experienced it now looks like users 
will need to put this snippet in their service_main:

# You will see this line in tdaemon.rb
while state == RUNNING || state == PAUSED || state == 0
   # Do stuff
end

If I remove the '|| state == 0' portion, I still experience deadlock 
while trying to stop the service.  I assume the 0 is the 
IDLE_CONTROL_CODE value that Patrick added.  Is that correct?

Also, I wonder if we could abstract that line down to something a little 
more user friendly.  Instead of this stock line:

while state == RUNNING || state == PAUSED || state == 0
   ...
end

How about abstracting that down to something like:

while running?
   ...
end

Just an idea.

Regards,

Dan






From phurley at gmail.com  Thu Oct 26 08:36:58 2006
From: phurley at gmail.com (Patrick Hurley)
Date: Thu, 26 Oct 2006 08:36:58 -0400
Subject: [Win32utils-devel] The win32-service patch is committed - and
	now, some questions
In-Reply-To: <45402EF5.6060105@gmail.com>
References: <45402EF5.6060105@gmail.com>
Message-ID: <554ac39c0610260536w6c804decna99c24454e7118c8@mail.gmail.com>

On 10/25/06, Daniel Berger <djberg96 at gmail.com> wrote:
> How about abstracting that down to something like:
>
> while running?


I like running? Not sure about the "0" state, as I have not looked at
the code for a while. I will take a peek, at the very least we should
provide a constant for the magic 0 state.

pth

From noreply at rubyforge.org  Thu Oct 19 10:12:16 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 19 Oct 2006 10:12:16 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6237 ] Add Guid API
Message-ID: <20061019141216.5135352417AC@rubyforge.org>

Patches item #6237, was opened at 2006-10-19 16:11
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6237&group_id=85

Category: None
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: Add Guid API

Initial Comment:
This adds API for creating GUIDs.

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6237&group_id=85

From noreply at rubyforge.org  Thu Oct 26 10:51:59 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 26 Oct 2006 10:51:59 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure ruby
	implementation
Message-ID: <20061026145201.3B93852414A4@rubyforge.org>

Patches item #6335, was opened at 2006-10-26 16:51
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85

Category: win32-shortcut
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: pure ruby implementation

Initial Comment:
This is (not finished) pure-ruby implementation of Win32::Shortcut.

Differences:
- new takes block (yield self at the end of initialize)
- open, relative_path, resolve and icon_number are not implemented
- instead of show_cmd window_style is used

Other than that, it should be pretty usable.

I've attached simple unit test as well

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85

From noreply at rubyforge.org  Thu Oct 19 10:11:53 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 19 Oct 2006 10:11:53 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6237 ] Add Guid API
Message-ID: <20061019141153.30F2952417A5@rubyforge.org>

Patches item #6237, was opened at 2006-10-19 16:11
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6237&group_id=85

Category: None
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: Add Guid API

Initial Comment:
This adds API for creating GUIDs.

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6237&group_id=85

From noreply at rubyforge.org  Thu Oct 19 10:10:46 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 19 Oct 2006 10:10:46 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6236 ] Add debug API
Message-ID: <20061019141046.69A12524177D@rubyforge.org>

Patches item #6236, was opened at 2006-10-19 16:10
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6236&group_id=85

Category: None
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: Add debug API

Initial Comment:
This adds couple of methods to interface with debugger (WinDbg, MS VC, etc.)

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6236&group_id=85

From noreply at rubyforge.org  Thu Oct 19 10:12:38 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 19 Oct 2006 10:12:38 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6236 ] Add debug API
Message-ID: <20061019141238.CA62252417B0@rubyforge.org>

Patches item #6236, was opened at 2006-10-19 16:10
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6236&group_id=85

Category: None
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: Add debug API

Initial Comment:
This adds couple of methods to interface with debugger (WinDbg, MS VC, etc.)

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6236&group_id=85

From noreply at rubyforge.org  Thu Oct 26 10:51:24 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Thu, 26 Oct 2006 10:51:24 -0400 (EDT)
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure ruby
	implementation
Message-ID: <20061026145125.2173F524141E@rubyforge.org>

Patches item #6335, was opened at 2006-10-26 16:51
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85

Category: win32-shortcut
Group: None
Status: Open
Resolution: None
Priority: 3
Submitted By: Jano Svitok (janek)
Assigned to: Nobody (None)
Summary: pure ruby implementation

Initial Comment:
This is (not finished) pure-ruby implementation of Win32::Shortcut.

Differences:
- new takes block (yield self at the end of initialize)
- open, relative_path, resolve and icon_number are not implemented
- instead of show_cmd window_style is used

Other than that, it should be pretty usable.

I've attached simple unit test as well

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85

From djberg96 at gmail.com  Fri Oct 27 20:14:09 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Fri, 27 Oct 2006 18:14:09 -0600
Subject: [Win32utils-devel] The win32-service patch is committed - and
	now, some questions
In-Reply-To: <554ac39c0610260536w6c804decna99c24454e7118c8@mail.gmail.com>
References: <45402EF5.6060105@gmail.com>
	<554ac39c0610260536w6c804decna99c24454e7118c8@mail.gmail.com>
Message-ID: <6037b70c0610271714w2baddf76p4c52b06be3e2eea9@mail.gmail.com>

On 10/26/06, Patrick Hurley <phurley at gmail.com> wrote:
>
> On 10/25/06, Daniel Berger <djberg96 at gmail.com> wrote:
> > How about abstracting that down to something like:
> >
> > while running?
>
>
> I like running? Not sure about the "0" state, as I have not looked at
> the code for a while. I will take a peek, at the very least we should
> provide a constant for the magic 0 state.
>
> pth


How about SERVICE_IDLE ?

Also, do we want to revisit RegisterServiceCtrlHandlerEx()?

- Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/win32utils-devel/attachments/20061027/f5282034/attachment.html 

From djberg96 at gmail.com  Fri Oct 27 22:57:10 2006
From: djberg96 at gmail.com (Daniel Berger)
Date: Fri, 27 Oct 2006 20:57:10 -0600
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure
	ruby	implementation
In-Reply-To: <20061026145201.3B93852414A4@rubyforge.org>
References: <20061026145201.3B93852414A4@rubyforge.org>
Message-ID: <4542C706.20303@gmail.com>

noreply at rubyforge.org wrote:
> Patches item #6335, was opened at 2006-10-26 16:51
> You can respond by visiting: 
> http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85
>
> Category: win32-shortcut
> Group: None
> Status: Open
> Resolution: None
> Priority: 3
> Submitted By: Jano Svitok (janek)
> Assigned to: Nobody (None)
> Summary: pure ruby implementation
>
> Initial Comment:
> This is (not finished) pure-ruby implementation of Win32::Shortcut.
>
> Differences:
> - new takes block (yield self at the end of initialize)
> - open, relative_path, resolve and icon_number are not implemented
> - instead of show_cmd window_style is used
>
> Other than that, it should be pretty usable.
>
> I've attached simple unit test as well
>   
What do folks think of using WScript.Shell like this?

- Dan

From phurley at gmail.com  Sat Oct 28 23:48:20 2006
From: phurley at gmail.com (Patrick Hurley)
Date: Sat, 28 Oct 2006 23:48:20 -0400
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure ruby
	implementation
In-Reply-To: <4542C706.20303@gmail.com>
References: <20061026145201.3B93852414A4@rubyforge.org>
	<4542C706.20303@gmail.com>
Message-ID: <554ac39c0610282048uef112e6j46a22a64cc1336d0@mail.gmail.com>

On 10/27/06, Daniel Berger <djberg96 at gmail.com> wrote:
> What do folks think of using WScript.Shell like this?

In general I love it -- the pure Ruby code is succinct and easy to
understand (as it lays on top of a high level api). My only concern is
that I have experienced difficult to reproduce hangs and odd behavior
(very rarely, but none the less), using WinOLE. I really only use it
against Excel and in my case it is not a big deal as it is easy to
stop and restart and things generally then preform as expected. Has
anyone else ever encountered this type of behavior? (or is it Excel
specific?)

pth

From Daniel.Berger at qwest.com  Mon Oct 30 08:32:21 2006
From: Daniel.Berger at qwest.com (Berger, Daniel)
Date: Mon, 30 Oct 2006 07:32:21 -0600
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure
	rubyimplementation
In-Reply-To: <554ac39c0610282048uef112e6j46a22a64cc1336d0@mail.gmail.com>
Message-ID: <39AA6550E5AA554AB1456707D6E5563D0259E8CA@QTOMAE2K3M01.AD.QINTRA.COM>

> -----Original Message-----
> From: win32utils-devel-bounces at rubyforge.org 
> [mailto:win32utils-devel-bounces at rubyforge.org] On Behalf Of 
> Patrick Hurley
> Sent: Saturday, October 28, 2006 9:48 PM
> To: Development and ideas for win32utils projects
> Subject: Re: [Win32utils-devel] [ win32utils-Patches-6335 ] 
> pure rubyimplementation
> 
> 
> On 10/27/06, Daniel Berger <djberg96 at gmail.com> wrote:
> > What do folks think of using WScript.Shell like this?
> 
> In general I love it -- the pure Ruby code is succinct and 
> easy to understand (as it lays on top of a high level api). 
> My only concern is that I have experienced difficult to 
> reproduce hangs and odd behavior (very rarely, but none the 
> less), using WinOLE. I really only use it against Excel and 
> in my case it is not a big deal as it is easy to stop and 
> restart and things generally then preform as expected. Has 
> anyone else ever encountered this type of behavior? (or is it Excel
> specific?)
> 
> pth

Although I haven't experienced the problems you describe personally, I
don't doubt you.  I know there are bugs in win32ole, and that library
could probably use a major refactoring.

If someone on this list ever felt like doing a rewrite of win32ole, I
certainly wouldn't mind.  I'd be happy to host it, too. :)

In the meantime, I've committed a pure Ruby version of win32-shortcut to
CVS (look for it in the toplevel directory, not under the win32utils
directory).  I took Jano's basic patch and expanded it to match the
current API, although there are some minor differences, mostly
improvements:

- The Shortcut.new method now takes a block, yields self, and
automatically calls Shortcut#save at the end of the block.
- The Shortcut#icon_number method is gone (not supported).  Pretty
useless method anyway, really.
- The Shortcut#hotkey method now takes a more human readable string
instead of a hex value.  Nicer.
- The Shortcut#show_cmd method was renamed Shortcut#window_style, as per
Jano's request (with an alias for b.c.)
- The Shortcut#window_style accessor now accepts/emits human readable
strings, as well as constants.
- The Shortcut.open method is more or less the same as Shortcut.new,
since pointing to an existing shortcut opens it by default - no special
code required.

Regards,

Dan


This communication is the property of Qwest and may contain confidential or
privileged information. Unauthorized use of this communication is strictly 
prohibited and may be unlawful.  If you have received this communication 
in error, please immediately notify the sender by reply e-mail and destroy 
all copies of the communication and any attachments.

From noreply at rubyforge.org  Mon Oct 30 23:58:21 2006
From: noreply at rubyforge.org (noreply at rubyforge.org)
Date: Mon, 30 Oct 2006 23:58:21 -0500 (EST)
Subject: [Win32utils-devel] [ win32utils-Patches-6335 ] pure ruby
	implementation
Message-ID: <20061031045822.2D020A970014@rubyforge.org>

Patches item #6335, was opened at 2006-10-26 07:51
You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85

Category: win32-shortcut
Group: None
>Status: Closed
>Resolution: Accepted
Priority: 3
Submitted By: Jano Svitok (janek)
>Assigned to: Daniel Berger (djberg96)
Summary: pure ruby implementation

Initial Comment:
This is (not finished) pure-ruby implementation of Win32::Shortcut.

Differences:
- new takes block (yield self at the end of initialize)
- open, relative_path, resolve and icon_number are not implemented
- instead of show_cmd window_style is used

Other than that, it should be pretty usable.

I've attached simple unit test as well

----------------------------------------------------------------------

>Comment By: Daniel Berger (djberg96)
Date: 2006-10-30 21:58

Message:
I've accepted the general idea in principle.  No need to submit any more patches for a bit - I've already got it done in CVS - unless you think I've missed something critical.

I'm going to put out a release tonight.  Feel free to submit any more patches or bug reports against the 0.2.0 release.

Many thanks,

Dan

----------------------------------------------------------------------

Comment By: Daniel Berger (djberg96)
Date: 2006-10-27 19:51

Message:
Thanks for this Jano.  It's definitely simple, but I need to discuss whether or not using Wscript.Shell is a good idea.

- Dan

----------------------------------------------------------------------

You can respond by visiting: 
http://rubyforge.org/tracker/?func=detail&atid=413&aid=6335&group_id=85