[xml] Sax and entity handling

[Rob Richards <rrichards ctindustries net>]

Would it be possible to consider reverting the behavior introduced in 
2.7 for the handling of predefined entities?

For the past month I have been trying to come up with a way to fix the 
xml extension in PHP, which breaks when used with 2.7.x, and have 
concluded that it is impossible to fix without having some sort of BC 
breakage or the order of entity processing changed back to its original 
state.

I understand that predefined entities can not be overriden, but by 
allowing getEntity to be called first, more flexibility is provided to 
the application layer.
For instance, within getEntity, I am handling the predefined entities by 
calling xmlGetPredefinedEntity so that the application is still 
compliant by not overriding any predefined entities. Why this is 
necessary is that the behavior of the sax parser doesn't follow exactly 
what libxml does, but rather is mimicking the behavior of expat. There 
are a number of differences, the biggest being expat's support for a 
default handler and difference in behavior whether this handler is 
defined or not.

In order to get all of this to work, within the PHP extension, all of 
the entity handling is performed within the getEntity call. To avoid 
having libxml perform any of the handling, the wellFormed flag is set to 
0, so within xmlParseReference, the function returns prior to calling 
any callbacks. Of course this is pretty much a hack, but was necessary 
to get the desired behavior.

That's pretty much it in a nutshell. I'm open to any other suggestions 
if going back to the old order is out of the questions.

Thanks,

Rob