Lists Home |
Date Index |
Elliotte Rusty Harold wrote:
>> I wouldn't have a problem, if SAX had always specified this, but
>> it didn't.
> One of the maintainers claims it did always specify this. One of the
> maintainers claims it didn't. Honestly, this is a mess; and sooner or
> later I think we should be pick one path or the other, but maybe not
> quite yet.
Unfortunately, I don't claim that SAX ever specified this clearly.
My original intention was that endDocument would not be called in an error
state, but I didn't write any docs to forbid doing so; David Brownell's
intention, as far as I can tell, was that endDocument would always be called
even in an error state, but he didn't remove my docs allowing it not to be
We do need to fix this problem, but we shouldn't do so gratuitously on a
minor maintenance release, since it's obviously a major compatibility
problem for both parsers and client code. Again, I think that the most
responsible thing to do is to flag the problem in the docs for both
endDocument() and fatalError(), so that people won't end up with a
(possibly-mistaken) assumption depending on which one they read first.
All the best,