[libxml-devel] Changing node equality

[Charlie Savage]

Dan Janowski wrote:
> I think this is a vestige of the old fact that there could be two ruby 
> objects that referred to the same XML node. Since I removed that and now 
> there is only a one-to-one relationship between the ruby peer and the 
> XML node, if == is equivalent to equal? and equal? is object identity, 
> then it makes sense to do as you say. Should eql? take on the code of 
> the current ==?

Good point, the new memory model definitely changes things.  Currently, 
the == operator in libxml is an alias for eql?, so there is only one method.

I did a bit of research to understand general ruby conventions.  The 
best write ups I found are:

http://groups.google.com/group/comp.lang.ruby/browse_thread/thread/572e8d8b01a7d24e/c5f532d868afab05
http://www.texperts.com/2007/10/16/navigating-the-equality-maze/

I like thinking about it this way (taken from one of the comments in the 
first post):

equal?       object identity test

   eql?       strict equality test, 1.eql?(1.0) is false
              (defaults to identity)

    ==        loose equality test, 1 == 1.0 is true
              (defaults to identity)

Usually == is the same as eql?, but not always (see the numeric classes 
for example).

In theory, you are supposed to override eql? for your own classes and 
perhaps ==.  If you do override eql? you are also supposed to override hash.

FOr libxml, I don't see how a custom eql? methods provides any benefits 
- why would you compare two xml nodes for equality, and if you did, what 
does that mean?  If the text content is the same, but the document is 
different, is that still equality?

The big downside is performance, all of a sudden code like this:

node1 == node2

becomes very slow.  In addition, lookups for hash tables or keys that 
contain nodes as keys also becomes slow.

So, I don't think defining eql? to do a text comparison while defining 
== to be object identity is a good idea because it would be inconsistent 
with existing ruby conventions and code.  So I propose we remove the 
overriden eql? method, thereby causing the == operator to default to an 
object identity test.

Charlie
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3237 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://rubyforge.org/pipermail/libxml-devel/attachments/20080707/63e4b0ed/attachment.bin>