This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: 3.3 problem of -fzero-initialized-in-bss
- From: "Joseph S. Myers" <jsm at polyomino dot org dot uk>
- To: Eric Botcazou <ebotcazou at libertysurf dot fr>
- Cc: Nathanael Nerode <neroden at twcny dot rr dot com>, gcc at gcc dot gnu dot org
- Date: Wed, 16 Jul 2003 00:26:30 +0100 (BST)
- Subject: Re: 3.3 problem of -fzero-initialized-in-bss
- References: <20030715220341.GA4593@twcny.rr.com> <200307160103.10072.ebotcazou@libertysurf.fr>
On Wed, 16 Jul 2003, Eric Botcazou wrote:
> I think that -fzero-initialized-in-bss should not have been made the default,
> the same way -fstrict-aliasing was not made the default when it was
> introduced.
You mean the way that it wasn't the default in EGCS 1.1, and people didn't
fix their software for it, and only after it had been enabled by default
in 2.95 and 2.95.1, then disabled in 2.95.2, did people fix their
software? (And then it was reenabled by default in 3.0, but the warning
option for broken code the 2.95.2 release notes vaguely offered didn't
appear until 3.3.) Not that it could particularly have been handled
better, but the first release with it disabled didn't get code fixed.
And "release with this feature turned on to get people to notice it, then
turn it off temporarily for code to be fixed" is hardly a strategy to
_plan_ to use, rather a response to an unforseen level of problems.
--
Joseph S. Myers
jsm@polyomino.org.uk