Henrik yes we are aware of that. Stef
On Feb 24, 2012, at 2:17 PM, Henrik Johansen wrote: > On Feb 24, 2012, at 1:35 PM, Nicolas Cellier wrote: > >> I systematically mark my fixes as ReviewNeeded because a second pair >> of eyes don't hurt. >> But this seems to be yet another bottleneck. >> Taking responsibility to review might seem scary and involving >> especially when changes touch obscure/complex part of the system. >> >> So maybe the review could take lighter, more or less automated forms: >> - does automated integration reports a regression ? >> - does the bug report gives enough technical solution details or a >> good rationale ? >> - are there some tests provided ? >> >> Maybe you could handle a confidence score per developer. with a priori >> and a posteriori informations. >> Like "good solution, but the integrator had more work to finish the job"... >> >> On the other hand, Pharo claims the right to do some mistakes (and >> soon correct them), so maybe this review phase could be relaxed a bit. >> >> And one thing I still ask for is a diff log for each SLICE posted in >> the inbox, entirely browsable from the issue tracker or from the bug >> list, because to review you need: >> - to read issue tracker report >> - AND to inspect the changes with Monticello merge in a recent clean image... >> Having the diff would be a great economy for the reviewers. >> Or maybe this already happened? >> >> Maybe the report generated by Monticello that pops up once the package >> commited, could contain such diff that we would manually copy/past in >> the bug entry. Would that be cheap enough ? >> >> Nicolas > > +999. > > I don't mind reviewing changes, but loading up an image, updating it to > latest version to ensure any conflicts are caught, opening MC browser, > finding the slice, diffing against image, and only THEN be able to do a > decent review are just too many manual steps, other than for issues I care > very strongly about. (in which case I'm often the one requesting ReviewNeeded > to begin with…) > > IMHO, doing all the above, and reporting merge conflicts/ new failed tests/a > diff to the issue would be a perfect task for Ulyssé :) > (As it is now, "Monkey went bananas" doesn't really tell me anything) > > Cheers, > Henry
