pantsbuild #general

Channels

welcome

github-notifications

announce

aloof-angle-91616

10/21/2019, 3:45 PM

not 100% sure -- it looks like it might have an undeclared setup-time dependency on pip? unless this is supposed to be working

cuddly-window-48195

10/21/2019, 3:46 PM

I think this is supposed to be working; I'm using the same setup as everyone else in my org, and pip is installed in my system environment?

cuddly-window-48195

10/21/2019, 3:47 PM

The package in question is a 3rdparty package via the 3rdparty pattern outlined in the documentation.

fast-author-5112

10/22/2019, 9:21 PM

pants question: when running

test.pytest

tests and looking at the coverage generated i see that several files are referenced through

.pants.d/pyrep/sources/

cache locations instead of their actual paths in our codebase. does anyone know why this happens? when combining coverage results for viewing in sonarqube all of those files are ignored with logs saying they no longer exist. any input would be appreciated

hundreds-father-404

10/26/2019, 7:55 PM

This error message UX is incredible: https://elm-lang.org/news/the-syntax-cliff Some possible places we could improve our error messages: - invalid CLI args (Danny did a great first pass at this) - invalid pants.ini - invalid BUILD files

👌 1

👖 1

❤️ 2

💥 1

aloof-angle-91616

10/26/2019, 8:24 PM

elm does a lot of really nice things and there’s a ton left to do with invalid CLI args (@fancy-motherboard-24956 and @dry-policeman-7927 were also looking for better ways to display help for cli args)

aloof-angle-91616

10/26/2019, 8:25 PM

Elm has a rule that any definition must be defined on a fresh line. It cannot have any spaces in front of it. One benefit of this rule is that the compiler can always pinpoint the particular definition that contains a syntax error. No more errors at the end of the file!

aloof-angle-91616

10/26/2019, 8:26 PM

luckily we’re not doing any parsing ourselves, parsing and typechecking error messages in general i think can be exactly as good or better than elm has, and a better parsing algorithm can actually help to make this a lot less effort

aloof-angle-91616

10/26/2019, 8:27 PM

one thing that’s true about the elm and rust compilers is that they both consider bad error messages to be a bug (and they therefore categorize and fix issues accordingly), which i would really really love it if we could adopt officially

aloof-angle-91616

10/26/2019, 8:29 PM

Elm uses capitalization to differentiate certain kinds of names. Module names start with an upper-case letter, like List and Http, whereas function and argument names start with a lower-case letter, like length and request. These rules make it easier to scan through new code, but they can be surprising in the learning phase, especially if you are used to JavaScript or English.

this is interesting and i actually like this idea in general, but i’ve had people go on about how using capitalization for specific identifiers is really confusing and a waste of a way for users to define their own meaning for caps

aloof-angle-91616

10/26/2019, 8:34 PM

thanks for sharing!!

enough-analyst-54434

10/26/2019, 11:58 PM

Style is totally subjective, consistency is not. I hope we can eventually provide a consistent experience in the positive sense. You may or may not like or grok Pants conventions initially, but you can trust them. This should be the bare minimum we strive for. We're not there yet.

👍 2

happy-kitchen-89482

10/28/2019, 11:16 PM

I'm noticing that the daemon isn't reloading the buildgraph when I edit requirements.txt. This was fixed over a year ago in https://github.com/pantsbuild/pants/issues/6404, but looks like it's now recurring. Can anyone think of a reason why it might?

happy-kitchen-89482

10/28/2019, 11:30 PM

huh, weird, now it's working again, it seems

salmon-barista-63163

10/30/2019, 7:11 PM

I’m trying to run pytests without reading or writing to the cache. Basically forcing all tests to run with every execution. I have tried adding:

Copy code

[cache.test.pytest]
cache_test_pytest_read: False
cache_test_pytest_write: False

to my pants.ini config but I am getting errors like this:

Copy code

19:10:16 [ERROR] Invalid option 'cache_test_pytest_write' under [cache.test.pytest] in /Users/briandesimone/workspace/talos/pants.ini
19:10:16 [ERROR] Invalid option 'no_cache_test_pytest_read' under [cache.test.pytest] in /Users/briandesimone/workspace/talos/pants.ini
timestamp: 2019-10-30T14:10:16.866302
Exception caught: (pants.option.config.ConfigValidationError) (backtrace omitted)
Exception message: Invalid config entries detected. See log for details on which entries to update or remove.

salmon-barista-63163

10/30/2019, 7:12 PM

Is there a better argument to pass to have tests not hit the cache?

witty-crayon-22786

10/30/2019, 7:13 PM

`--cache-ignore`: can see a related discussion in the #jvm room

witty-crayon-22786

10/30/2019, 7:15 PM

with regard to `'cache_test_pytest_write' under [cache.test.pytest]`: once the option has been scoped by the section, it would just be `write`: so `'write' under [cache.test.pytest]`: ie,

Copy code

[cache.test.pytest]
read: False
write: False

salmon-barista-63163

10/30/2019, 7:15 PM

@witty-crayon-22786 thank you thats what i needed.

witty-crayon-22786

10/30/2019, 7:15 PM

can see https://www.pantsbuild.org/options.html#setting-option-values for more info.

calm-artist-46894

10/31/2019, 9:11 AM

I am investigating an import issue with my pex file: in a virtual env or a clean environment my pex run fine; but it doesn't in my regular development environment. After some investigation, it turns out I have an older version of jinja2 (2.8) installed in

/usr/local/lib/python2.7/dist-packages/

and my application requires 2.10. Funny thing is before any of my code is executed, the python runtime already loaded a bunch modules, noticeably all

jinja2

sub-modules but not

jinaj2

itself. Then when my code is importing

jinja2

it would try to load some submodules and eventually fails with module not found error because the new version of

jinja2

was looking for some new functions in its sub-modules.

calm-artist-46894

10/31/2019, 9:20 AM

Did another test with

repl

Copy code

xxx@yyy:~/zzz$ ./pants repl zirak/src/python/example:hello2
Python 2.7.15rc1 (default, Nov 12 2018, 14:31:15)
Type "copyright", "credits" or "license" for more information.

IPython 1.0.0 -- An enhanced Interactive Python.
?         -> Introduction and overview of IPython's features.
%quickref -> Quick reference.
help      -> Python's own help system.
object?   -> Details about 'object', use 'object??' for extra details.

In [1]: import sys
In [2]: [k for k in sys.modules.keys() if "jinja2" in k]
Out[2]:
['jinja2.collections',
 'jinja2.cPickle',
 'jinja2.errno',
 'jinja2.lexer',
 'jinja2.threading',
 'jinja2.compiler',
...

Question is how does these modules get populated?

calm-artist-46894

10/31/2019, 8:00 PM

Is there a similar slack channel for

pex

related issue?

happy-kitchen-89482

10/31/2019, 8:13 PM

#pex

🙏 1

wide-energy-11069

11/01/2019, 6:02 PM

Is pants-devel moderated? email didn’t seem to show up at https://groups.google.com/forum/?hl=en#!forum/pants-devel

witty-crayon-22786

11/01/2019, 6:06 PM

yes, it is

witty-crayon-22786

11/01/2019, 6:07 PM

first time sending from that email address?

witty-crayon-22786

11/01/2019, 6:07 PM

oh. i see it...?

damp-easter-17791

11/02/2019, 6:00 PM

Is there a way to compile transitive dependencies only given a target? For example

Copy code

./pants compile --transitive-only src::

This would compile everything except the targets that match

src::

damp-easter-17791

11/02/2019, 6:04 PM

Alternatively, is there a way to pass in a large argument list via text file? If I compute the list of transitive dependencies and run

./pants compile $TRANSITIVE_TARGETS

then the argument is too long