[nomerge][2.12] Welcome back Partest!#9169
Conversation
9ccb0f7 to
fe12d64
Compare
This comment has been minimized.
This comment has been minimized.
fe12d64 to
4cdabef
Compare
4cdabef to
3d12db1
Compare
|
It's quite some pandemic busy work. I'm surprised there is any forward merging to 2.13, and I would be surprised if partest maintenance is worth the efforts. OTOH, repairing the Vulpix split might make maintenance easier for 2.13/3.0. I have tinkered with dotty but I've forgotten again how to run a test. |
|
@sjrd will this affect Scala.js? |
|
It probably will, yes. But we're used to juggling with different versions of partest depending on the version of Scala, so nothing disruptive I think. |
|
I might have to comment on the youtube video, Anyone else here because of Welcome Back, Partest? By coincidence, recently I've punned on Kotter when working in Kotlin. |
lrytz
left a comment
There was a problem hiding this comment.
👍
I'm surprised there is any forward merging to 2.13
We got bitten a few times by the mismatch. At least flags are not silently ignored in 2.13, but for backports the situation is still unsafe (/* scalac: -Werror */ is silently ignored). And migrating between the two formats is really annoying.
|
I'm surprised there is any backporting to 2.12, but I can see how that would be annoying. I'll look into a Scalafix for partests! |
3d12db1 to
6651375
Compare
|
proposal to archive the old repo: scala/scala-partest#119 |
The motivation is to bring "/* scalac: ... */" pragma support to 2.12.x's partest, as easily as possible, as backporting/forward-merging changes is frequently impacted by 2.12's lack of pragma support and 2.13's lack of flags-file support...
So, as an alternative to scala/scala-partest#118, here's a without-history c+p of 2.13.x's partest, for the 2.12.x branch.
I decided to defer backporting the "testkit" split out and other related build tweaks that followed partest's re-integration into the 2.13.x branch (though I might've copied some details), particularly I had been looking at:
I also decided to defer both migrating all the flags files to pragmas as well as dropping flags support (which I actually had to restore - all the other post-copy tweaks were just mechanical changes, like scala.util.chaining and JavaConverters.)
I've tried to make a reasonable attempt at putting together a workable change, without sinking too much time into this.
But feedback welcome, particular by @som-snytt. Specifically could you check the
ExtConsoleReporter/DirectCompilerclose/finish/hasError logic, please?