Re: [xml] Potential wrong usage of xmlIsID() in tree.c

[Rob Richards <rrichards ctindustries net>]

Kasimier Buchcik wrote:
> Hi,
>
> Bugzilla:
> http://bugzilla.gnome.org/show_bug.cgi?id=331273
>
> Regards,
>
> Kasimier
>   
Kasimier,

To give you an idea of how I think the IDs should be modified, take a 
look at this diff:
http://www.ctindustries.net/libxml/tree.c.diff.txt

It's still a work in progress, as I am also fixing up some of the other 
sibling functions. These aren't DOM compliant functions (DOM only works 
on children with these functions), but the library in general has always 
allowed attributes to be added using these functions (as they are 
generic and not DOM specific), so I don't think they should prevent it 
now and probably should deal with IDs as well. There are a few potential 
crashes dealing with attribute additions in these which is what I'm 
currently still fixing.

As far as copy/import/adopt goes. I have found a couple of conflicting 
opinions for this. One is that the importing doc should determine if an 
attribute is added as an ID. The other (which is also how it is 
implemented in Xerces) is that an attribute should be copied/imported 
with the same ID-ness as in the source document. The latter would be in 
line with how XInclude works and imo would make it easy for a user to 
retrieve by ID in the importing doc.

I also found out that all attributes require an element to be considered 
an ID (even the manually set ones), so modified the code appropriately 
for this. In the event the copy/import goes the way of the importing doc 
determining ID-ness, the manually set IDs do copy over as IDs regardless 
of what the importing doc says, which would add a bit more complexity on 
top of the XInclude handling.

Thoughts?

Rob