adjust tests on JDK 17+ for PermittedSubclasses change

[SethTisue]

addresses the test failures seen at e.g. https://github.com/scala/scala/actions/runs/4792344656

references #10105

[SethTisue]

@som-snytt this is an expected progression, I hope?

it wasn't obvious to me if it's possible to ask for the -old test to run on 8 and 11, rather than just 8. there don't seem to be any examples of syntax for that in the repo currently.

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 17+ here could be made more accurate, I'd be happy to adjust it

[som-snytt]

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.

[SethTisue]

it's only necessary to specify javaVersion once per round

yeah I wondered about that but figured the redundant comments wouldn't hurt 🤷

[som-snytt]

yep, only my feelings. Sticks and stones, but not redundant comments.

[som-snytt]

I totally missed your comment and question about javaVersion. I swear I just weaked it to accept spaces in the range

javaVersion: 8 - 11

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 t8700 because I only partest --branch now.

Edit: it was the opposite, I removed the spaces, apparently because the line is always split now, for some important reason.

60394a9

The PR comment explains

Note "javaVersion" is only used as a lower bound, "19+", as in dotty, the thought leader, so for simplicity [sic] a range must be written without spaces, "11-19".

I forgot to adjust my test path:

sbt:root> partest neg/t8700b
Discarding 1 invalid test paths
Selected 0 tests drawn from the well of despair. I see you're not in a testing mood.

So actually javaVersion can be written just once for all test files. (But the new caching makes these checks less painful, too.) Verifying that range still works:

-- 1 - neg/t8700b-old                            [skipped on Java 19, only running on 8-11]

[som-snytt]

TIL why this regressed.

scala/bug#12800

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.