On Tue, Jun 28, 2011 at 3:20 PM, Rainer M Krug <r.m.k...@gmail.com> wrote: >> This all makes sense to me. I would only insist in grouping >> lyx-a.b-svn with lyx-a.b-trunk. As much as latest branch would usually >> be more stable than the latest stable release, the branch could easily >> contain nasty regressions: it is still by definition unreleased code. >> We wouldn't want to encourage people using lyx-a.b-svn over lyx. And >> this way we can have a PPA for daily builds and one for releases. But >> that's possibly just my preference. > > I agree with you, but it was stated earlier in this discussion, that > BRANCH_2_0_X is as stable (possibly even more stable) than the actual last > release - so it should not be together with daily-trunc. But I think you are > absolutely right: to put it into the same ppa with the releases, would > indicate an "always stable" daily build, which it is not. > Yes, it may be more stable than latest stable, but it might not. There is a reason that sometimes we have regressions in latest stable releases.
> One option would > be to trigger, whenever the BRANCH is considered as stable, a manual build > is triggered (which goes into lyx-release), while the daily buids still go > into lyx-daily. Now the question would be who declares BRANCH as stable for > this stable-branch build? > I don't think we need to go into such an intricate set-up, at least because providing so many builds is already a boon while providing too many can get very confusing. The way I see it, branch is considered stable when we opt for a new stable release. At all costs we don't want regular pre-releases for the stable branch: it's like releasing experimental code for our stable code. I think daily lyx-a.b-svn builds are just fine for the purpose, which is providing people that are having some sort of nastiness in the latest stable release with bleeding-edge code of branch that fixes the issue. This should be the only encouraged use of branch daily builds, apart from testing. Cheers Liviu
