Unfortunately, regular branch builds fail too often, and with false negatives, to trust them.
Nexus transient issues apart (is there an INFRA ticket?), this is one of the side effects of having so many repos. Downstream builds need the main repo snapshots properly populated and most failures are due to those builds running "too early", before those are ready. Most failures are from the downstream repos and caused by this. If we can workaround it in some way, I' m sure we'll considerably reduce the noise and false negatives. That said, having those builds, personally, gives me the confidence for the commit-then-review practice. In this case I'm pretty sure the scope of the version bump wasn't known (I personally don't remember most of the details in the blobstore providers) and understand that the use of the OkHttp can be unknown for some (fair enough!). Fortunately we have the CI builds (when fixed) and individuals reviewing the commit notificarions, and for me that's enough. Commit-then-review will always have the counterpart that someone else could have something to say to the change, that will always be there, but that is part of the "then-review" part. This exactly what we did here by replying to the commit notification mentioning some additional checks. IMO the process we have works and I personally trust every committer. I am quite comfortable with it and believe that the current process and tools give us enough confidence for the commit-then-review practice. El 19/06/2014 22:54, "Jeremy Daggett" <jeremy.dagg...@rackspace.com> escribió: > FWIWŠ Another data point for the discussion here. > > Over the past couple of weeks, I started working on a PR submitted for the > OpenStack Glance v2 > API (thanks @sallum): > https://github.com/jclouds/jclouds-labs-openstack/pull/77 > > I updated to 2.0.0-RC2 as soon as it was released to the masses. My intent > is to use the 2.0.0 version by > default as soon as it is officially released. > > /jd > > On 6/19/14, 12:53 PM, "Andrew Gaul" <g...@apache.org> wrote: > > >Everyone will have their own opinion on this, but raising the bar with > >consensus tends to lower the velocity of development and quantity of > >contributions. Gaining consensus for a dependency upgrade is far below > >my bar for notifying others or desiring to be notified. If a given > >commit causes a regression, we can always revert it. > > > >On Thu, Jun 19, 2014 at 07:06:42PM +0000, Everett Toews wrote: > >> Agreed that commit-then-review is fine for smaller changes but this > >>doesn¹t fall into that category. > >> > >> The okhttp.version affects downstream modules (e.g. openstack-marconi) > >>and changing that version could break those downstream modules. I don¹t > >>classify that as a small change. > >> > >> Also, committing directly to the ASF git repo circumvents the CI jobs > >>when a pull request is made in GitHub. Those jobs could catch any > >>obvious problems and provide a layer of protection from breaking the > >>build. > >> > >> I completely understand the desire to move fast on some of this stuff. > >>It would be nice if you could give those of us who are maintaining > >>downstream modules a heads up on IRC or via email. > >> > >> Regards, > >> Everett > >> > >> > >> On Jun 19, 2014, at 12:30 PM, Andrew Gaul <g...@apache.org> wrote: > >> > >> > Committers are free to use commit-then-review or review-then-commit. > >>I > >> > often use the former for smaller changes. > >> > > >> > On Thu, Jun 19, 2014 at 02:48:24PM +0000, Everett Toews wrote: > >> >> Where¹s the pull request for this? > >> >> > >> >> Everett > >> >> > >> >> > >> >> On Jun 18, 2014, at 9:50 PM, g...@apache.org wrote: > >> >> > >> >>> Repository: jclouds > >> >>> Updated Branches: > >> >>> refs/heads/master 1982626cd -> ad56db2a4 > >> >>> > >> >>> > >> >>> Upgrade to OkHttp 1.6.0 > >> >>> > >> >>> This upgrades both drivers/okhttp and mockwebserver. Changelog: > >> >>> > >> >>> https://github.com/square/okhttp/blob/master/CHANGELOG.md > >> >>> > >> >>> > >> >>> Project: http://git-wip-us.apache.org/repos/asf/jclouds/repo > >> >>> Commit: > >>http://git-wip-us.apache.org/repos/asf/jclouds/commit/ad56db2a > >> >>> Tree: http://git-wip-us.apache.org/repos/asf/jclouds/tree/ad56db2a > >> >>> Diff: http://git-wip-us.apache.org/repos/asf/jclouds/diff/ad56db2a > >> >>> > >> >>> Branch: refs/heads/master > >> >>> Commit: ad56db2a4d6ac4c3b7b126ba4303ab6e0c1f8b5e > >> >>> Parents: 1982626 > >> >>> Author: Andrew Gaul <g...@apache.org> > >> >>> Authored: Wed Jun 18 19:41:46 2014 -0700 > >> >>> Committer: Andrew Gaul <g...@apache.org> > >> >>> Committed: Wed Jun 18 19:49:50 2014 -0700 > >> >>> > >> >>> > >>---------------------------------------------------------------------- > >> >>> project/pom.xml | 2 +- > >> >>> 1 file changed, 1 insertion(+), 1 deletion(-) > >> >>> > >>---------------------------------------------------------------------- > >> >>> > >> >>> > >> >>> > >>http://git-wip-us.apache.org/repos/asf/jclouds/blob/ad56db2a/project/pom > . > >>xml > >> >>> > >>---------------------------------------------------------------------- > >> >>> diff --git a/project/pom.xml b/project/pom.xml > >> >>> index 2d3b651..4b259d0 100644 > >> >>> --- a/project/pom.xml > >> >>> +++ b/project/pom.xml > >> >>> @@ -201,7 +201,7 @@ > >> >>> > >><jclouds.test.listener>org.jclouds.test.testng.UnitTestStatusListener</jc > >>louds.test.listener> > >> >>> <test.ssh.keyfile /> > >> >>> > >><sourceReleaseAssemblyDescriptor>source-release-zip-tar</sourceReleaseAss > >>emblyDescriptor> > >> >>> - <okhttp.version>1.3.0</okhttp.version> > >> >>> + <okhttp.version>1.6.0</okhttp.version> > >> >>> <surefire.version>2.17</surefire.version> > >> >>> </properties> > >> >>> > >> >>> > >> > > >> > -- > >> > Andrew Gaul > >> > http://gaul.org/ > > > >-- > >Andrew Gaul > >http://gaul.org/ > >
