I'm not familiar enough with the testing infrastructure scripts to review the details of these changes in the PR [1] (is there someone more familiar that could review?), but internal-only pre-commit and full scan post-commit seems like a good plan to me. From staring at a lot of link testing logs for awhile now, most real issues are internal (usually when pages are added/moved) and these are good to catch during PR review. Dead external links are fairly rare, so I agree a weekly review of post-commit results for actual 404s would probably be sufficient.
[1] https://github.com/apache/beam/pull/8318 On Tue, Apr 16, 2019 at 11:38 AM Kyle Weaver <kcwea...@google.com> wrote: > > it would be good to have a sort of weekly report on dead links > > Seeing as checking for broken external links returns a lot of false > positives, I'd rather not spam everyone with them. However, I don't > know if making it a postcommit will give it sufficient visibility. Not > sure what the best way to handle this is--only thing I am sure of is I > don't want this blocking PRs. > > Kyle Weaver ️ Software Engineer ️ github.com/ibzib ️ > kcwea...@google.com ️ +16502035555 > > > On Tue, Apr 16, 2019 at 1:13 AM Ismaël Mejía <ieme...@gmail.com> wrote: > > > > +1 to removing link validation for website changes. However it would be > good to have a sort of weekly report on dead links or another alternative > to be aware of them. > > > > On Tue, Apr 16, 2019 at 2:43 AM Kyle Weaver <kcwea...@google.com> wrote: > >> > >> I agree with Andrew that the external links checks are ultra-flaky and > seldom strictly needed, so I filed a PR to make checking external links > optional and disabled by default: https://github.com/apache/beam/pull/8318. > Let me know what you all think. > >> > >> Kyle Weaver ️ Software Engineer ️ github.com/ibzib ️ > kcwea...@google.com ️ +16502035555 > >> > >> > >> > >> On Mon, Apr 1, 2019 at 11:05 AM Kenneth Knowles <k...@apache.org> > wrote: > >>> > >>> +1 thanks for noticing and raising yet another source of > non-hermeticity (plus the docker constraint) > >>> > >>> On Mon, Apr 1, 2019 at 9:09 AM Andrew Pilloud <apill...@google.com> > wrote: > >>>> > >>>> +1 on this, particularly removing the dead link checker from default > tests. It is effectively testing that ~20 random websites are up. I wonder > if there is a way to limit it to locally testing links within the beam site? > >>>> > >>>> On Mon, Apr 1, 2019 at 3:54 AM Michael Luckey <adude3...@gmail.com> > wrote: > >>>>> > >>>>> Hi, > >>>>> > >>>>> after playing around with Gradle build for a while, I would like to > suggest to remove ':beam-website:testWebsite target from Gradle's check > task. > >>>>> > >>>>> Rationale: > >>>>> - the task seems to be very flaky. In fact, I always need to add '-x > :beam-website:testWebsite' to my build [1] > >>>>> - task uses docker, which imho adds a (unnecessary) severe > constraint on the build task. E.g. A part time user is unable to execute > these tests in a docker environment > >>>>> - these tests are accessing production environment. So myself > hitting the build several times an hour could be considered a DOS attack. > >>>>> > >>>>> Of course, these tests add lots of value and should definitely be > executed, but wouldn't it be sufficient, to run this task only dedicated, > i.e. by an explicit call to ':beam-website:testWebsite' o > ':websitePreCommit'? Any thoughts? > >>>>> > >>>>> best, > >>>>> > >>>>> michel > >>>>> > >>>>> [1] https://issues.apache.org/jira/browse/BEAM-6760 >
