but actually...

for options you want on all targets, you just want

--compile-zinc-args

. you're likely seeing that warning because you changed the value of

--compile-zinc-whitelisted-args="{'key1':val1,'key2':val2,...}" (default: { '-S.*':'False','-C.*':'False','-file-filter':'True','-msg-filter':'True' })
    A dict of option regexes that make up pants' supported API for zinc. Options
    not listed here are subject to change/removal. The value of the dict
    indicates that an option accepts an argument.

perhaps?

oh. or because that's just an invalid arg, and so it's not matching the regex

hmm, yes, I had to wrap with double-quotes for some reason, but possibly not really.

Another weird error that seemingly shouldn't happen:

Exception in thread NailgunStreamWriter:
                     Traceback (most recent call last):
                       File "/Users/benjyw/.pyenv/versions/3.6.6/lib/python3.6/threading.py", line 916, in _bootstrap_inner
                         self.run()
                       File "/Users/benjyw/.cache/pants/setup/bootstrap-Darwin-x86_64/1.17.0.dev1_py36/lib/python3.6/site-packages/pants/java/nailgun_io.py", line 271, in run
                         self._in_fds.remove(fileno)
                     ValueError: list.remove(x): x not in list

If you look at that code (https://github.com/pantsbuild/pants/blob/805ef8a2887872cf8851b68ac00fad6ff282be97/src/python/pants/java/nailgun_io.py#L245)

errored

is supposed to be a subset of

self._in_fds

Could there be dups in

errored

?

that's with pantsd, yea?

@average-vr-56795 was mentioning seeing this error a while ago^ cc @red-balloon-89377

yeah

We can work around it by squelching the

ValueError

, but it would be nice to know why it's happening.

The

select()

syscall modifies the argument lists in-place, is it possible Python does the same? The documentation implies not, but doesn't explicitly say so.

I need some general python

mock

help, if someone has a minute: I have something like:

fake_environment = {"PANTSD_REQUEST_TIMEOUT_LIMIT": "-1"}
    self.mock_handler_inst.parsed_request = lambda: (None, None, [], fake_environment)

,

which is called from different threads

If I print the result of

handler.parsed_request()

from the threads, I get:

# The first thread
ENv is {'PANTSD_REQUEST_TIMEOUT_LIMIT': '-1'}
# The second thread
ENv is {}

Why do I get no output in the second one?

hi, is there a task/option to find out which target fetch a given third party dependency ?

I'm playing with

resolve.ivy

but it's kind of tedious right now

With a recent pants version update some coworkers and I encountered this error:

Traceback (most recent call last):
  File "/home/alexis/.cache/pants/setup/bootstrap-Linux-x86_64/virtualenv-16.4.3/virtualenv.py", line 22, in <module>
    import distutils.spawn
ModuleNotFoundError: No module named 'distutils.spawn'
./pants: line 170: /home/alexis/.cache/pants/setup/bootstrap-Linux-x86_64/pants.fPKtJQ/install/bin/pip: No such file or directory
./pants: line 171: /home/alexis/.cache/pants/setup/bootstrap-Linux-x86_64/pants.fPKtJQ/install/bin/pip: No such file or directory
New virtual environment successfully created at /home/alexis/.cache/pants/setup/bootstrap-Linux-x86_64/1.15.0_py36.
./pants: line 193: /home/alexis/.cache/pants/setup/bootstrap-Linux-x86_64/1.15.0_py36/bin/python: No such file or directory

After some investigatiosn it appeared with

1.15.0.dev0

version

i was able to fix that when i saw it yesterday by running

pip3 install setuptools==40.0.0

, which i didn't try to push a fix for because i thought it was isolated to the environment i was in, and it was running pants from source -- could you let me know if that does anything?

Hello good people! I am back and continue to work on Zipkin tracing in Pants. Zipkin tracing didn’t work properly with multiple threads and I am in the process of changing it. Thus I have a question: Background workunits in Pants do not have a parent workunit. So currently in Pants at least two workunits do not have a parent. But in Zipkin the span without a parent id is considered the root of the trace. So Zipkin trace that has two spans without parent in Zipkin UI looks very bad. Is it fine for

background_root_workunit

to be a child of the main workunit? Pictures show how the trace will look like in this case. If this solution is not very good I would be grateful for better one 🙂

@witty-crayon-22786 @average-vr-56795

Yeah, I think it’s fine for background to be a child of main… We could also maybe give it a different span type (“pants background workunit” rather than “pants workunit” maybe)?

Who wants a one-character code review? https://github.com/pantsbuild/pants/pull/7820

i agree with this!

i learned today these behaviors of pants options that surprised me: (1)

./pants test --compile-zinc-cache-ignore

works -- i would have assumed that this would become

--test-compile-zinc-cache-ignore

, which doesn't exist (2)

./pants test --junit-cache-ignore

fails -- i would have assumed this would become

--test-junit-cache-ignore

, analogous to

./pants test.junit --cache-ignore

, except additionally: (3)

./pants --test-junit-cache-ignore

isn't recognized!! resulting in

Exception message: Unrecognized command line flags on scope 'test.junit': --cache-ignore.

this is despite the fact that

./pants test.junit --cache-ignore

works!! i think these feel like fixable bugs -- one approach is e.g. to "normalize" all command lines into options with scopes, e.g. convert

test.junit --cache-ignore

into

--test-junit-cache-ignore

, and then modify the code that looks for options to be able to recognize that (it currently doesn't -- that's #3 above). has anyone experienced any other surprises with options? does anyone have any thoughts on whether changing the options behavior to fix 1, 2, 3 above might break existing behavior (e.g. in scripts calling pants)?

Unrelated options question… How can I from one task inspect another task’s options?

Can I get a rubber stamp for https://github.com/pantsbuild/pants/pull/7852?

Pants_build_fails_with_DistUtilsError_Cython_.m