-
Notifications
You must be signed in to change notification settings - Fork 0
Conversation
e1eac7a
to
cd68250
Compare
Wim, I've resolved the conflict in pom.xml. Do you plan to continue here or should I complete this one? |
Go ahead! Let me know if you need help. |
Introduce clean target Replace site.xml with category.xml Make org.eclipse.mylyn.commons-parent self-sufficient Upgrade findbugs to spotbugs Signed-off-by: Alexander Fedorov <[email protected]>
Align org.hamcrest.core Signed-off-by: Alexander Fedorov <[email protected]>
@wimjongman I advanced to test launch, but it currently fails |
Try to configure surefire Signed-off-by: Alexander Fedorov <[email protected]>
Fix several unit tests Signed-off-by: Alexander Fedorov <[email protected]>
Neutralize requirement for org.hamcrest Signed-off-by: Alexander Fedorov <[email protected]>
Add dependabot Signed-off-by: Alexander Fedorov <[email protected]>
Revert changes in tests done by bulk replace http->https Signed-off-by: Alexander Fedorov <[email protected]>
Switch to required-bundle to simplify configuration Signed-off-by: Alexander Fedorov <[email protected]>
Revert changes in tests done by bulk replace http->https Signed-off-by: Alexander Fedorov <[email protected]>
Change expected Java version range to 11.0-18.0 Signed-off-by: Alexander Fedorov <[email protected]>
…g.eclipse.mylyn.commons.git into Releng_for_e4.26_#8
Revert changes in tests done by bulk replace http->https Signed-off-by: Alexander Fedorov <[email protected]>
@ruspl-afed, @wimjongman: We actual have the targets in two places.
So maybe we should look that we can
|
Well, it will mean having a lot of excessive items and longer fetch each time. Is there any gain in this strategy?
Do you have a suggestion how to organize such a "reuse" technically? @BeckerFrank it depends on the decision are we going to merge git repos or not.
|
Revert changes done by bulk replace http->https Fix FeedReader tests Signed-off-by: Alexander Fedorov <[email protected]>
@wimjongman after many hours I've found why it is so tricky to get the test working: the bulk replace "http"->"https". |
Revert changes done by bulk replace http->https Fix org.eclipse.mylyn.commons.repositories.http.tests Signed-off-by: Alexander Fedorov <[email protected]>
@BeckerFrank I'm going to merge this one, since it enables CI Please create an issue to discuss our target platform strategy |
I'm sorry I wasted your time. I did not consider the tests, I just tried to get the build going. I saw that the targets were using http, the features, the update sites, and the license URL, so I figured let's throw a regex at it. 🙏 |
Yes, I saw that. The mylyn repo is the pom parent of the commons repo. This is a lot of releng to juggle around. I agree with Alexander that the repos should be self-sufficient. However, it is also the case that some repos have to be built before others. common itself is in use by other downstream repositories. This requires a cascading build setup: When commons is built, it must trigger its downstream builds. We can also look at commons (and other upstream repos) as just another third-party dependency. It means that downstream repositories are built with whatever common is currently in the snapshot/nightly repo. When a release is planned, we can make one build that cascades all builds into proper order and then release that result. |
With my Platform experience having too many repos is an issue - both for releng and for contributors so it's worth first asking which repos deserve to be kept separate before spending time to make them selfsufficient. |
We are discussing this here: #14 |
BTW currently we have circular dependency between org.eclipse.mylyn.monitor.* and org.eclipse.mylyn.context.*, to be resolved later #16 .
I fully agree here, one need to think carefully before splitting. However, we with @BeckerFrank identified a potential use case to have at least one Mylyn repo separately - to build/install it for older versions of Java/Eclipse. Please join our discussion #14 |
Move all http to https
Create a new target file