Re: [tree-ssa] dead const/pure/alloca call removal

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Andreas Schwab <schwab@suse.de> writes:

[...]

| > | > | I disagree that the C standard's definition is too weak to allow
| > | > | eliding of calls, as-if.  What problems do you see?
| > | >
| > | > The as-fi would apply *if* GCC is the malloc implementation provider.
| > | > Currently , it is not,
| > | 
| > | There is no way for a strictly conforming program to notice the elision.
| >
| > How many useful programs are strictly conforming?
| 
| That's all that you can depend on when you are arguing with the standard.

No. The C standard also talks about conforming programs. But if you
see previous mails of mine, I've been arguing against making taking
that slipery slope. 

FYI, this malloc/free thingy has been postulated as being mostly
beneficial for C++ programs, but C++ has no notion of strictly
conforming programs. 

| > And it the thing that is of interest is to provide a compiler for
| > strictly conforming programs, then "cp /bin/sh gcc" is a far cheaper
| > answer. 
| >
| > | All others must resort to extensions and/or implementation details.
| >
| > Yes.  But "strict conformance" is not all about it.
| 
| But the standard refers to the implementation's documentation when it
| comes to non-strictly-compliant behaviour.  GCC is part of the
| implementation.

Are you proposing that GCC should rule out tracing allocators?  That
would be a great step toward making the compiler  useless, just in the
name of speed.

-- Gaby