https://pantsbuild.org/ logo
#development
Title
# development
a

ancient-vegetable-10556

03/23/2022, 6:18 PM
Is there a way to get processes to echo their stdout/stderr automatically, or do we need to explicitly log their captured output? (@witty-crayon-22786)
w

witty-crayon-22786

03/23/2022, 6:18 PM
the latter.
a

ancient-vegetable-10556

03/23/2022, 6:18 PM
Oooft.
w

witty-crayon-22786

03/23/2022, 6:19 PM
are you looking at the JVM output issue in 2.11.x?
a

ancient-vegetable-10556

03/23/2022, 6:19 PM
Yeah. Trying to get that to happen systematically, without repeating the output for a single source file if it happens to be a repeated dependency
w

witty-crayon-22786

03/23/2022, 6:19 PM
are you seeing the lack of output for all compiles, or only for some?
a

ancient-vegetable-10556

03/23/2022, 6:19 PM
All compiles
a

ancient-vegetable-10556

03/23/2022, 6:19 PM
there’s no code to output the errors at all
yes, it is
w

witty-crayon-22786

03/23/2022, 6:20 PM
that code is supposed to raise the
level
of the workunit for the compile to
ERROR
, and then render the message
a

ancient-vegetable-10556

03/23/2022, 6:20 PM
Well, it looks like the relevant rust code only prints the first line of the message
w

witty-crayon-22786

03/23/2022, 6:20 PM
is there an ERROR line, but no message?
got it.
ok, sec.
(fwiw: it’s possible that bisecting would be worthwhile… but let me get you a pointer to the relevant codepath)
oh… actually. since it’s rendering the ERROR, that means that at least part of the EngineAwareReturnType mechanism is working.
can add a
Copy code
log::warn!("...")
in there to see whether it is being called at all, and then go from there
but since the level is actually being adjusted to ERROR, that implies that
fn level
in the same file is actually taking affect.
a

ancient-vegetable-10556

03/23/2022, 6:27 PM
let me poke around for just a moment
w

witty-crayon-22786

03/23/2022, 6:31 PM
also: feel free to open an issue for this one and assign it to me… i suspect that it is related to my workunit changes, rather than something at a higher level
(particularly if
30640b2225
is the change that introduced the issue)
a

ancient-vegetable-10556

03/23/2022, 6:32 PM
So I get this (from
JvmProcess -> FallibleProcessResult
)
Copy code
11:28:37.84 [INFO] Completed: Compile testprojects/src/jvm/org/pantsbuild/example/lib/ExampleLib.java with javac
And then there’s no subsequent logs related to the
FallibleClasspathEntry
I’ll see if bisecting might help
w

witty-crayon-22786

03/23/2022, 6:33 PM
you mentioned that there was an ERROR line?
but yea, i’m fairly sure that this is workunits. i can take a look
a

ancient-vegetable-10556

03/23/2022, 6:34 PM
I think I may have been confused with the
Completed
line,
Let me, at the very least, bisect to confirm for you
w

witty-crayon-22786

03/23/2022, 6:35 PM
k. i would go so far as suggesting confirming before/after
30640b2225
a

ancient-vegetable-10556

03/23/2022, 6:35 PM
since I have a reproductioncase here
(words!)
@witty-crayon-22786 confirmed.
javac
output present on 0aa7f290ed390ab3ceba0c44e4b38354d7a41d3d but not 30640b2225f7e8efd302ffa4d7f771b933570ca7
w

witty-crayon-22786

03/23/2022, 6:41 PM
ok, thanks. i’ll tackle this one in a little bit.
a

ancient-vegetable-10556

03/23/2022, 6:42 PM
I’ve added a commit that shows a minimal reproduction case that doesn’t get caught by javaparser etc (which do output issues)