On Mon, Sep 11, 2017 at 7:19 AM, Stephane Ducasse <stepharo.self@gmail.com> wrote:
Hi torsten
From a process I think that we decided that external packages should be managed like the following:
- PR (from people) to change the code in Pharo + issues a PR to Iceberg.
- then Iceberg team merge PR + issue a new PR for Pharo integration
So I think that this is what you describe too?
Yes, that's what I described in ESUG too :) https://www.slideshare.net/GuillePolito1/understanding-the-pharo-dev-process... On the way back from ESUG we discussed with Esteban a way to support this kind of scenarios cleanly and without needing too much of a refactor ;)
Like that we do not have deadlock and can always have a stable version.
Stef
On Sun, Sep 10, 2017 at 7:16 PM, Torsten Bergmann <astares@gmx.de> wrote:
Hi,
to explain:
we have lots of uncategorized methods and non commented classes. For some time we had to accept this ugly situation that old and legacy code. For Pharo 4 and 5 I invested a lot of time to clean this up right before the release date. But with 6.0 such a round was not done and in 6.0 and 6.1 it looks like a mess again.
Why? Because we introduced new features and code and never defined a certain level of quality for code we include. But our initial goal with Pharo was (and hopefully still is) to cleanup things up and do better than before.
So we should not forget about quality and so I spend some time last on this again - now for Pharo 7. For instance #setUp and #tearDown in SUnit should be in "running" as it was defined for TestCase subclasses. Also #hash and #= should be in "comparing" protocol. Also the goal is coming back to an image where all methods are categorized and where all classes have a comment.
But for me it absolutely makes no sense to reinvest the time over and over into the same cleanups while others can easily make a mess again.
So we should RAISE THE QUALITY BAR to ensure that we keep with such a quality level. Especially as see each build to be a release and the real release once a year often is more or less a snapshot.
So I also wrote a new test called "ProperMethodCategorizationTest" and while cleaning up #hash it was green. So ProperMethodCategorizationTest. testHashMethodNeedsToBeInComparingProtocol showed no problem and I committed and have sent a Pull request. This looked like any other new contribution.
What I did not knew at this time was that Iceberg code is not in git repository - but managed externally. So while I fixed the code in my image with wrongly categorized hash method in an Iceberg class (IceSemanticVersion) the system did not show me that this change did not move to GitHub.
Now I know - but this has such bad side effect when we do changes and these packages are in the image - but not managed with the pharo repository.
Long story short: one can not fix this situation with a simple PR as Iceberg code is not under the "pharo" repo umbrella.
I already submitted a PR for Iceberg
https://github.com/pharo-vcs/iceberg/pull/458
and Esteban included that already - but only in Iceberg. We now need to include Iceberg.
Several lessons learned from my side: - doing changes and having green tests in the image is not enough - also doing a commit and PR does not mean all of your changes are on GitHub - it is not a good situation when a part of the image code is managed with "pharo" repository and another part is managed externally - I see this as a problem of the new process we should discuss and address - we should nonetheless try to define tests to show the edges where we can improve and cleanup (without a test it would not have been noticed that while I did my best to cleanup the current way of managing Iceberg prevented my fix to be into the image in the first place)
I dont know what need to be done to include a new iceberg version or when this will happen from Estebans side - but as soon as it is done this test should be green again.
Thanks T.
Gesendet: Sonntag, 10. September 2017 um 18:24 Uhr Von: "Stephane Ducasse" <stepharo.self@gmail.com> An: "Pharo Development List" <pharo-dev@lists.pharo.org> Betreff: Re: [Pharo-dev] argh, tests are failing!
Yes we know but have no idea how to recategorise Iceberg protocols
On Sun, Sep 10, 2017 at 1:03 PM, Pavel Krivanek < pavel.krivanek@gmail.com[mailto:pavel.krivanek@gmail.com]> wrote:
Of course you are right, the Integrators should really take care on it.
Before it were only common random failures but the main problem is with this PR: https://github.com/pharo-project/pharo/pull/264[https:/ /github.com/pharo-project/pharo/pull/264]
The new test testHashMethodNeedsToBeInComparingProtocol fails on classIceSemanticVersion
-- Pavel
2017-09-10 10:13 GMT+02:00 Guillermo Polito <guillermopolito@gmail.com[ mailto:guillermopolito@gmail.com]>: Hi all,
Since a couple of builds we have consistently failing the following tests:
GT.EventRecorder.Tests.Core.GTEventRecorderTest.testDeliverNow2[ https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull% 20request%20and%20branch%20Pipeline/job/development/ 110/testReport/junit/GT.EventRecorder.Tests.Core/GTEventRecorderTest/ testDeliverNow2/]GT.EventRecorder.Tests.Core.GTEventRecorderTest. testNotDeliveredDataShouldBeResent[https://ci.inria.fr/ pharo-ci-jenkins2/job/Test%20pending%20pull%20request% 20and%20branch%20Pipeline/job/development/110/testReport/ junit/GT.EventRecorder.Tests.Core/GTEventRecorderTest/ testNotDeliveredDataShouldBeResent/]Kernel.Tests.Processes.MutexTest. testFailedCriticalSectionShouldUnblockWaitingOne[https://ci. inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull% 20request%20and%20branch%20Pipeline/job/development/ 110/testReport/junit/Kernel.Tests.Processes/MutexTest/ testFailedCriticalSectionShouldUnblockWaitingOne/] ReleaseTests.Categorization.ProperMethodCategorizationTest. testHashMethodNeedsToBeInComparingProtocol[https://ci.inria. fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request% 20and%20branch%20Pipeline/job/development/110/testReport/ junit/ReleaseTests.Categorization/ProperMethodCategorizationTest/ testHashMethodNeedsToBeInComparingProtocol/]ReleaseTests.Categorization. ProperMethodCategorizationTest.testHashMethodNeedsToBeInCompa ringProtocol[https://ci.inria.fr/pharo-ci-jenkins2/job/Test% 20pending%20pull%20request%20and%20branch%20Pipeline/job/ development/110/testReport/junit/ReleaseTests.Categorization/ ProperMethodCategorizationTest/testHashMethodNeedsToBeInCompa ringProtocol_2/]ReleaseTests.Categorization.ProperMethodCategorizationTest .testHashMethodNeedsToBeInComparingProtocol[https://ci.inria. fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request% 20and%20branch%20Pipeline/job/development/110/testReport/ junit/ReleaseTests.Categorization/ProperMethodCategorizationTest/ testHashMethodNeedsToBeInComparingProtocol_3/]ReleaseTests.Categorization. ProperMethodCategorizationTest.testHashMethodNeedsToBeInCompa ringProtocol[https://ci.inria.fr/pharo-ci-jenkins2/job/Test% 20pending%20pull%20request%20and%20branch%20Pipeline/job/ development/110/testReport/junit/ReleaseTests.Categorization/ ProperMethodCategorizationTest/testHashMethodNeedsToBeInCompa ringProtocol_4/]ReleaseTests.Categorization.ProperMethodCategorizationTest .testHashMethodNeedsToBeInComparingProtocol[https://ci.inria. fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request% 20and%20branch%20Pipeline/job/development/110/testReport/ junit/ReleaseTests.Categorization/ProperMethodCategorizationTest/ testHashMethodNeedsToBeInComparingProtocol_5/]ReleaseTests.Categorization. ProperMethodCategorizationTest.testHashMethodNeedsToBeInCompa ringProtocol[https://ci.inria.fr/pharo-ci-jenkins2/job/Test% 20pending%20pull%20request%20and%20branch%20Pipeline/job/ development/110/testReport/junit/ReleaseTests.Categorization/ ProperMethodCategorizationTest/testHashMethodNeedsToBeInCompa ringProtocol_6/]ReleaseTests.Categorization.ProperMethodCategorizationTest .testHashMethodNeedsToBeInComparingProtocol[https://ci.inria. fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request% 20and%20branch%20Pipeline/job/development/110/testReport/ junit/ReleaseTests.Categorization/ProperMethodCategorizationTest/ testHashMethodNeedsToBeInComparingProtocol_7/]
Green builds are the only hard metric to say if the build is healthy or not.
- We should not integrate anything until the build is green again...
Also, we spent with Pablo a lot of time to have a green build in all platforms... I'd like to spend my time in other fun stuff than the CI :/
--
Guille Polito
Research Engineer French National Center for Scientific Research - http://www.cnrs.fr[ http://www.cnrs.fr]
Web: http://guillep.github.io[http://guillep.github.io] Phone: +33 06 52 70 66 13
-- Guille Polito Research Engineer French National Center for Scientific Research - *http://www.cnrs.fr* <http://www.cnrs.fr> *Web:* *http://guillep.github.io* <http://guillep.github.io> *Phone: *+33 06 52 70 66 13