Everything

php

BC fix for __call definition. (Bug #29716) + pear bug #2136

php.internals

Alan Knowles

22 years ago

Attached (hopefully) should be a fix for bug #29716, allowing 3
arguments for __call, but emitting a E_STRICT error.

This is needed to enable simple BC wrappers to be written for code
wanting to use overload in PHP4.

BTW: Notes on current CVS

flex 2.5.31 currently reports
flex: fatal internal error, bad line in skeleton file
flex 2.5.4: works OK (debian package flex-old)

/ext/standard/unserializer.c needs touching, otherwise re2c tries to
generate it (and fails here)

Zeev/Andi could you OK it?, marcus is ok with the principle (although he
hasnt seen the code)

Regards
Alan

Andi Gutmans

22 years ago

I actually don't think it's the right approach. ext/overload has always
been experimental and it doesn't make sense to start putting such a hack
into the engine.
You can easily change the code to use conditional includes and therefore,
your PHP 4 version won't be compiled under PHP 5.
Do you know how many PEAR classes are actually affected by this?

Andi

At 08:58 PM 8/19/2004 +0800, Alan Knowles wrote:

Alan Knowles

22 years ago

Andi Gutmans wrote:

> I actually don't think it's the right approach. ext/overload has
> always been experimental and it doesn't make sense to start putting
> such a hack into the engine.

Normally that would make sense, but the change is realitively minor, and
it would reduce the necessity to upgrade libraries for alot of users.

> You can easily change the code to use conditional includes and
> therefore, your PHP 4 version won't be compiled under PHP 5.

I guess so this may end up being the only option, other than kludgy
evals, it's just a bit of a shame to have to include a file for  just
for 5 lines of code.

> Do you know how many PEAR classes are actually affected by this?

2 PEAR classes XML_FastCreate and DB_DataObject.
Regards
Alan

Andi Gutmans

22 years ago

At 10:29 AM 8/20/2004 +0800, Alan Knowles wrote:

>Andi Gutmans wrote:
>
>>I actually don't think it's the right approach. ext/overload has always
>>been experimental and it doesn't make sense to start putting such a hack
>>into the engine.
>
>Normally that would make sense, but the change is realitively minor, and
>it would reduce the necessity to upgrade libraries for alot of users.
>
>>You can easily change the code to use conditional includes and therefore,
>>your PHP 4 version won't be compiled under PHP 5.
>
>I guess so this may end up being the only option, other than kludgy evals,
>it's just a bit of a shame to have to include a file for  just for 5 lines
>of code.
>
>>Do you know how many PEAR classes are actually affected by this?
>
>2 PEAR classes XML_FastCreate and DB_DataObject.

Two are very few. I don't think eval's with heredocs would be that kludgy.
At least not for such a small amount of classes.

Andi