'Re: "Re: level 1 core - hc_nodecloneattributescopied fails"'

on 4/7/2004 3:40 AM Andrew Clover wrote:
Hi,

> Kasimier Buchcik <kbuchcik@4commerce.de> wrote:
> 
> 
>>This test fails the assertion of equality on my side, since the first 
>>attribute name is "title" and the first collection item has a value of 
>>"class".
> 
> 
> Yes, that's why 'expected' is of type Collection rather than List. When 
> compared with a Collection, order is irrelevant.

Now I see! I never though about the difference between a Collection and 
a List.

> (This was the assumption I made for the Python test runner, anyway - I 
> haven't seen any definitive TSML docs on this.)
> 
> 
>>I guess that cloning of an element does not imply any change of 
>>the position of the attributes
> 
> 
> Attributes do not have a position (fixed ordering) in the DOM spec in 
> any case, so it would be acceptable (if not generally likely) for an 
> implementation to return different item(0) values for a NamedNodeMap and 
> its clone.

Yes, and not having thought well enough about the semantic of 
Collections and Lists, I started to doubt about the sanity of the tests. 
Well, I'll try to trust them more in the future.

>>I just started to change the test-to-java stylesheet to produce Delphi code
> 
> 
> Phew! That sounds fun! ;-) Let us know how you get on...

It seems to be a relatively smooth port until now, since we use 
Interfaces for all the DOM constructs; if I had explicitely to "free" 
every instance created, it would need a lot of additional work on the 
stylesheet.


Thanks!

Kasimier

Received on Wednesday, 7 April 2004 05:17:33 UTC