This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Warnings in the C++ Front-End and GCC in General
- To: james dot g dot mansion at hsbcgroup dot com
- Subject: Re: Warnings in the C++ Front-End and GCC in General
- From: Craig Burley <burley at gnu dot org>
- Date: Fri, 11 Sep 1998 09:39:49 -0400 (EDT)
- Cc: egcs at cygnus dot com
- Cc: burley at gnu dot org
>I want to disable warnings *in certain contexts* only.
This is exactly one of my points, which never really got addressed,
probably because the facility was withdrawn. It's why we probably
won't ever offer a general-purpose facility to disable warnings --
too many people will insist it be applicable to very narrowly
defined constructions, and for good reason, while making that work
would take a substantial *design* effort, not to mention implementation
and maintenance effort.
I'd rather not discuss these requirements in detail anymore on this
list, since they're clearly well beyond the scope of whatever work
people want to do. If someone wants to create a detailed requirements
document and post the URL here, however, that might be welcome. It
should address issues such as "exactly where, in the source, must
source text to disable an uninitialized-variable warning be present
to actually disable that warning -- where the variable is declared,
in the given instance where the variable might be referenced without
having been clearly defined prior to that point, where the variable
is implicitly declared (e.g. via Fortran's IMPLICIT), or in some
precise combination of these?".
As I've suggested in my previous emails, it is my belief that, if
this issue was "formally" addressed using the classic model whereby
requirements were first agreed upon, then design, then implementation,
it would not get further than partway before the requirements phase
before people would realize just how much work -- short- and long-term
-- this feature really is. Managing number/name assignments is indeed
a drop in the bucket.
tq vm, (Burley)