On 25/10/11 17:39 -0400, Matt Wagner wrote:
On Tue, Oct 25, 2011 at 01:12:09PM -0700, Richard Su wrote:
> I think would should have a policy whereby people who are checking in
> code to master should be monitoring the the builds to ensure they pass.
> In the case where someone forgets to include a file in a commit, I would
> say, if the author is online, the author should have a chance to fix the
> failure. If the author is unavailable then I think, the code should be
> reverted immediately.
I agree 100% here. If the developer that broke tests can fix them right
then and there, I'm all for that. But in "I'll look into that in a
little bit" cases, we should just revert the code until it's fixed.
_______________________________________________
I think reverting code is a bad direction here. Not only is it
irritating for the commiter to find out something that was ACKed has
been reverted, but it make for messy git history. Not that we can
never do it, but I think it shoudl be used very sparingly. Unless it
is a complete blocker for anyone else _immediately_, I think the
offending developer should be notified (maybe via jenkins?in the
meantime on list by whoever discovers the breakage) that they have
broken the build. It then becomes their TOP PRIORITY to correct their
error. Any in-progress task should be set aside (committed in their
topic branch, switch to a new one to fix the bug), until the build is
fixed. Having this happen to you once or twice will make you much
more careful to check that your have not broken the build, imo - and
it allows you to correct things without having your work reverted.
-j