This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: builtin_apply et al
- From: Stan Shebs <shebs at apple dot com>
- To: Nicola Pero <nicola at brainstorm dot co dot uk>
- Cc: Nathan Sidwell <nathan at codesourcery dot com>, gcc at gcc dot gnu dot org
- Date: Fri, 16 Apr 2004 16:09:00 -0700
- Subject: Re: builtin_apply et al
- References: <Pine.LNX.4.44.0404162342400.29145-100000@nicola.brainstorm.co.uk>
Nicola Pero wrote:
I can see that the builtin_apply *might* be useful somewhere, but is it
really being used beyond test cases?
Dare I suggest deprecating this 'feature'? (rather than fixing it)
The consensus among the few who've thought about this is that
yes, it should flushed from libobjc and then officially deprecated.
Hmmm. :-)
Yes/maybe ... but what are you suggesting to replace it with ?
Some sort of reliable and portable mechanism must be provided by the
compiler (or by compiler libraries) to replace it, because the Objective-C
libraries do need a way to do such stuff.
I've been told in the past it should be replaced with libffi - the main
problem there is ... how much reliable and portable is libffi ?
Needed for Java, so gets regular attention for that reason, and has
platform specific code for lots of platforms. Libffi could be thought
of as an implementation of builtin_apply.
ObjC needs a certain amount of visibility into C calling conventions
to do its thing, but it's tricky to generalize and not many hackers
work on libobjc (the NeXT/Apple runtime mostly does it with assembly
code, and is thus not very portable).
Not sure what your point is here.
Just calibrating expectations - it may be a while before builtin_apply
can go away.
Stan