-
Notifications
You must be signed in to change notification settings - Fork 49
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable Github Verification and cleanup build scripts #237
Conversation
Test Results 6 files 6 suites 1h 9m 26s ⏱️ For more details on these failures, see this check. Results for commit 6c993ae. ♻️ This comment has been updated with latest results. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK, it happened what I've expected.
As long as github actions fail and reporting is as bad as it is, I would not merge this.
I personally have hard time to understand the "fail" output, and which build produced which "error" (which is not an error but test fail, but that's known issue).
I don't want / don't have time to explain every contributor on every PR that the various fails that we report are to be ignored (!!!), because later it causes that even experienced committer assume the PR is "safe" just because it fails everywhere else.
See for example eclipse-platform/eclipse.platform.ui#703 (comment)
=>
The way to proceed is to analyze fails and why they differ between maven / jenkins / official builds and to provide patches that would fix test fails or at least mute the known issues on github only.
a quick look seem to indicate the usual problem with no default JRE found on mac os I'll try to add a p2.inf for that ... as mentioned elsewhere failures are not to be ignored I have never claimed that nor is it recommended or fine in any way as it show for example that currently people on macos won't be able to execute the maven build, just as we ignored such fact in the past and said "don't worry if the (linux) build succeed" does not mean that's good in terms of code quality and build stability. |
The problem is, as long as there are multiple fails, one has to ignore them otherwise no PR in platform would be allowed to be merged. So people learn to ignore "the noise", and that is even worse as not to have build results on other platforms. |
I don't agree on the implication that people has to ignore these, even if they do. Everyone can first help stabilize the test/builds before going on with new features (and some even do so) but as long as these failures are not obvious it is very unlikely that one can effectively doing this. The main problem is that people classify anything that gets in their way as "noise" instead of useful pointers that something is wrong and should be fixed beforehand or at least documented (!). Still there is a risk that this hides bugs in new implementations as it might adds up to the failure that currently is already revealed. |
63ceb7d
to
3cd1174
Compare
You can agree or disagree, but the reality is that the github actions are failing since they introduction and everyone ignores the fails and merges happily to master. I don't want this state in JDT =>
|
That is simply not true, platform has 1(!) failing test on mac in now because people are helping stabilize the failing tests already, so all linux and all windows test are green now ... so just because some people ignores them do not mean everyone ignores these. So its the choice of the project or the developer to ignore them, its not a natural law or "reality" ... |
See eclipse-platform/eclipse.platform.ui#721 (comment). I don't want this state in JDT =>
|
What should I see? That people just thing they could ignore checks? As said your project lead so your task is to lead the team to follow project rule, if the rule is "ignore any errors because we don't care" that's what happening ... and would have happened without github builds as well, as people have ignored freeze period (first by accident, then by intend). |
alright this already fixed a lot of the failing one, lets see whats next... |
Seems now a JRE is found but not the right one: test requires a JVM >= 9
@iloveeclipse do you have an idea what can make macOS fragment think this is desired? Actually we make Java 8, 11, 17 (default) available but it chooses the Java 8 as default. |
Just leave 17 please.
I have no idea how and who resolves what on macOS. |
I believe it was needed because of new API's provided by Java 19 |
Would you please rebase so we can see current status here? |
3cd1174
to
977495b
Compare
Done. |
This do the follwoing: - enable the unified github verification build - removes obsolete workflows - move mandatory maven options into the maven.config file - add p2.inf to pull in fragment for macos launches
977495b
to
6c993ae
Compare
@laeubi Do you plan to finish this one soon or it should be closed? |
Actually it is more the choice of JDT team... I can only offer the setup but I can'T debug possible failures or decide if its usefull for JDT or not. |
As contributor can not work on getting clean build, PL said without it it's a no-go and no one else stepped up to do anything in year and a half it's obvious this is not going anywhere. Closing. |
This do the follwoing:
FYI @iloveeclipse I intentionally left out the
-Ptest-on-javase-19
on the Github build for now because I assume this is not mandatory?