adjust tests on JDK 17+ for PermittedSubclasses change
#10383
Conversation
|
@som-snytt this is an expected progression, I hope? it wasn't obvious to me if it's possible to ask for the this would still be mergeable as-is I think, but I'd be happy to tweak it, too it came up recently that I like to pretend that JDK 9–10/12–16/18–19 don't exist, but if you happen to know that the |
PermittedSubclasses change
There was a problem hiding this comment.
Thanks, my apologies, I saw this failure at some point but must have switched branches and cleaned. Please feel free to ping me to do my own follow-ups, I almost said foul-ups.
For future ref, I learned that in partest compilation rounds, "no number" comes first, so for the common case, C.java doesn't need an awkward number, then s_2.scala (or any number >= 0) is second. Also, // scalac tool arg comments are merged per round, so it's only necessary to specify javaVersion once per round; by definition, the files are compiled together.
Soon, partest knowledge will be only of historical interest.
yeah I wondered about that but figured the redundant comments wouldn't hurt 🤷 |
|
yep, only my feelings. Sticks and stones, but not redundant comments. |
|
I totally missed your comment and question about I just spent a half-hour trying to find where, but to no avail. It must be somewhere. The upside to a slow PR queue is that later, I'll notice the feature go that's so cool. I bet I forgot about Edit: it was the opposite, I removed the spaces, apparently because the line is always The PR comment explains
I forgot to adjust my test path: So actually |
|
TIL why this regressed. Sometimes I look back at my squandered youth and think what an ignorant wastrel I was! Other times, I regard not the Springtime of my Youth but just last spring with the same amazed disdain. |
addresses the test failures seen at e.g. https://github.com/scala/scala/actions/runs/4792344656
references #10105