Re: [Gnu-arch-users] GCC v. Arch address@hidden: Regressions on mainline]

[Tobias C. Rittweiler]

On Wednesday, June 23, 2004 at 4:14:53 AM, 
    Colin Walters <address@hidden> wrote:

> On Tue, 2004-06-22 at 19:15 -0700, Tom Lord wrote:
>
> > GCC commits happen too fast (last I checked) to serialize them while
> > inserting tests between each one.
>
> How else would you do it?  You could run the tests in parallel, which
> would scale up faster with more CPUs, but if you want to enforce the
> invariant that the test suite passes for every commit on the mainline -
> you have to run the test suite for every commit.

Don't test on *every* commit, but firstly pile those [commits] up. Just
from time to time do a build-test-cycle:


As Tom nicely pointed out, in a project assumed to be rapidly changing
it's practically very likely for anyone who checks out a version of
this project that she does not have the most recent version at times
after the checkout, build or hackery anyway.

Because of that it's not of much matter for a developer that the
mainline is akin a hydra that she cannot get ahead [pun intended] of.
Rather the matter for her is how to get a buildable and reasonably
usable version, methinks.

And the provision of such a version should be the outcome from
*periodically* running a build-test-cycle like once a day which will
either declare the current revision at test to be the "most recent
version verified for buildability" or just ignore it.

For a GCC-like project it probably would not matter whether or not the
build-test-cycle of one day succeeds, but after a constant failure
throughout a longer interval (say 3 weeks) the PQM should probably send
some notification to responsible persons.


-- tcr (address@hidden)  ``Ho chresim'eidos uch ho poll'eidos sophos''