David M. Lloyd
David M. Lloyd
I think we may be missing the forest for the trees here: is this more than a matter of filtering out `module-info.java` when we call `javac` in dev mode?
Exactly, the only difference here is that projects with `module-info` descriptors would stop failing in dev mode per this bug report. There would be no change in our requirements, or...
If building or deployment does fail with JPMS, then please open a separate issue for this case. The fix for this issue would not affect that one way or the...
~~Could you try out #36200 and let me know if that fixes the issue for you @rteabeault and/or @kpagratis and/or @serahl and/or @y-luis-rojo?~~ See @cescoffier's fix instead, please!
I think this was more or less implemented in an alternative branch, at least the important aspects.
Not sure if this is relevant here, but concurrency (particularly concurrent startup) can affect the "high tide" in ways that can be hard to predict.
We are not likely going to use this solution. Discussion is ongoing regarding a potential bootstrap API which would be the replacement. It's not fully decided yet though.
The failure now seems to have something to do with a socket timeout related to TLS: ``` 2024-06-14T21:19:28.5820922Z [ERROR] Tests run: 2, Failures: 0, Errors: 1, Skipped: 0, Time elapsed:...
Same problem again. :( It looks like there could be a possible way for invalid URLs to be created on Windows, so at least one possible explanation would be that...
@dependabot rebase