This is the mail archive of the
mailing list for the GCC project.
Re: GCC 4.0 Status Report (2005-02-03)
- From: David Edelsohn <dje at watson dot ibm dot com>
- To: Steven Bosscher <steven at gcc dot gnu dot org>
- Cc: gcc at gcc dot gnu dot org, Mark Mitchell <mark at codesourcery dot com>
- Date: Tue, 08 Feb 2005 20:51:44 -0500
- Subject: Re: GCC 4.0 Status Report (2005-02-03)
- References: <firstname.lastname@example.org>
>>>>> Steven Bosscher writes:
> Given the net rate of regression fixing, and the magnitude of the
> changes in 4.0, expecting a _shorter_ time between branching and
> releasing seems a bit strange to me.
> Perhaps branching at 100 regressions is just too early?
Counter arguments include:
1) We have released previous versions of GCC without every regression
fixed. Not all regressions are show-stoppers.
2) The types of regressions has changed as users are testing more code and
more corner cases. Some of the C++ regressions were present in GCC 3.4
and the world survived through GCC 3.4.3 so far.
3) Because more optimization infrastructure is based on Tree-SSA, more of
the compiler is common code. Instead of debugging strange
target-dependent interactions with the RTL optimizers and ambiguities in
RTL itself, many problems now are reproducible on all targets, which helps
with the bug fixing process.