[Boost-testing] Regression policy for mainline?

From: AlisdairM (alisdair.meredith_at_[hidden])
Date: 2007-01-26 03:32:26

I have been working on a package of Borland fixes for the mainline
(been avoiding RC for obvious reasons) which means I have a large
number of uncommitted patches in my source tree.

I have been working on the priciple that I cannot post the patches
without some demonstration they work and don't break things, so have
set up a regular regression run from my CVS checkout (mainline only)

This has been running since early January, but I have not yet seen
results since mainline page generation froze on January 5th - which is
a few days before my first results were posted so I hope is unrelated.

The new process is talking about locking down configs, which sounds
like I would lose my option to test my patches, which is difficult to
work with as I have not yet got an offline viewer to even verify for
myself which regressions show up (currently doing this by hand, which I
am sure is unreliable)

While I understand RC comes first, I want to be clear I will be able to
run and publish tests on HEAD when the current problems are resolved.

(I also plan to add Cygwin runs for Doug's ConceptGCC and Variadic
Template compiler once the Borland runs are stable - this will be
clearly marked as an 'experimental' runner)

-- 
AlisdairM