blargh... subsystem global state has got to go.

would be a huge change to have to pass instances everywhere, but the sideeffecting initialization is just too painful.

Rookie question: I should make a fork of the repo, clone that, and then make branches for each new bug I work on, correct?

Correct

See: http://www.pantsbuild.org/howto_contribute.html 'Getting Pants Source Code'

That explains the fork part at least

@happy-kitchen-89482 I'm seeing this testing out pants 1.2.0 in our repo: UninitializedSubsystemError: Subsystem "Arguments" not initialized for scope "target-arguments". Is subsystem missing from subsystem_dependencies() in a task?

Looks like this was touched in the last commit before the 1.2.0 release

see https://rbcommons.com/s/twitter/r/4303/ prolly (EDIT: or, add explicit sources arguments to targets)

Is this in a custom task?

I mean one of your own tests, rather than pantsbuild tests?

I've noticed this when there's some other error, and the error-handling code somewhere (I forget where) expects that Argument is set up, so the original error gets masked... 😞

but yea, if it's in a test, it's expected when you don't specify

sources

to targets, or when that subsystem isn't initialized and you try to create a target

This is in a test of a custom task so yes on boht accounts.

@ancient-stone-50795 : the edits made to tests in https://rbcommons.com/s/twitter/r/4300/diff/6#index_header are the relevant bit then... either initialize the subsystem, or give targets empty sources lists with

sources=[]

@witty-crayon-22786 I made a similar edit a little while ago in our repo and the test now passes

thx

👍

@brash-oxygen-26029 is getting an error trying to work in pants master:

~/Development/pants master ./pants help-all --verbose
    Updating registry `<https://github.com/rust-lang/crates.io-index>`
error: An unknown error occurred

and I think I just ran into something similar:

./pants --help-all
A rust installation could not be found, installing via the instructions at <https://www.rustup.rs> ...
info: downloading installer
info: syncing channel updates for 'stable-x86_64-apple-darwin'
info: downloading component 'rustc'
 41.2 MiB /  41.2 MiB (100 %)   3.4 MiB/s ETA:   0 s                
info: downloading component 'rust-std'
 54.1 MiB /  54.1 MiB (100 %)   3.4 MiB/s ETA:   0 s                
info: downloading component 'rust-docs'
  6.7 MiB /   6.7 MiB (100 %)   3.4 MiB/s ETA:   0 s                
info: downloading component 'cargo'
info: installing component 'rustc'
info: installing component 'rust-std'
info: installing component 'rust-docs'
info: installing component 'cargo'
info: default toolchain set to 'stable'

  stable installed - rustc 1.12.1 (d4f39402a 2016-10-19)

info: using existing install for 'stable-x86_64-apple-darwin'
info: override toolchain for '/Users/zundel/Src/pants-release' set to 'stable-x86_64-apple-darwin'

  stable-x86_64-apple-darwin unchanged - rustc 1.12.1 (d4f39402a 2016-10-19)

    Updating registry `<https://github.com/rust-lang/crates.io-index>`
 Downloading fnv v1.0.3
error: unable to get packages from source

To learn more, run the command again with --verbose.
~/src/pants-release master

yep, when I retry I get the same output as Chris

do you have access to github from these machines?

and it's possible that it is flakiness... if you try a few times, does it work?

i think @fancy-queen-20734 and @billowy-beach-68498 saw similar flakiness, but it stabilized

@witty-crayon-22786 yes, we do have access to github, this was immediately preceeded by a

git pull

from the pants repo. We're both getting it consistently

(Just tried again)

Actually I didn't see flakiness at all. I guess what I did differently was I downloaded rust installer online and install rust before I did git pull and pants command

hm... that shouldn't be necessary. i uninstalled and reinstalled it a few times, and confirmed that it should work without having been manually installed.

cc @enough-analyst-54434

...oho

@User uploaded a file: