From: Douglas Gregor (gregod_at_[hidden])
Date: 2002-02-09 12:15:08
On Friday 08 February 2002 01:36 pm, you wrote:
> Does anyone have an opinion about the lifetime of our release candidate
> branches? My feeling is that < 1 week isn't long enough, and that we should
> give it a bit more time before releasing it. Opinions?
I'd like a least a week before we release, and would also like some guideline
for the release manager to follow. I also think that the guideline should be
in some way related to regression testing.
Should every test be required to run correctly on every compiler? Surely not,
because most compilers are terrible and that's not a reasonable burden. I
think we should select a single platform that is the closest to correct, and
require that all regression tests pass on that compiler prior to a release.
The closest I've come to compiler/platform nirvana is Comeau 184.108.40.206 or GCC
3.0.3 on Linux using a recent version of glibc for wide character support,
and we're actually doing quite well on those platforms now.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk