pantsbuild #general

Channels

welcome

github-notifications

announce

witty-crayon-22786

11/09/2015, 3:29 AM

sure

ancient-stone-50795

11/09/2015, 3:30 AM

Another solution would be if we could set up some kind of target relationship and say that if this target changes, incremental compile on the java library is out the window. That's essentially what we tried to do with our custom task/target.

witty-crayon-22786

11/09/2015, 3:30 AM

yea, that's likely what we'll need to do

ancient-stone-50795

11/09/2015, 3:31 AM

You guys have a similar case?

witty-crayon-22786

11/09/2015, 3:31 AM

zinc doesn't invalidate for things like removed classpath entries it seems

ancient-stone-50795

11/09/2015, 3:31 AM

Oh, I see how that could make a difference.

witty-crayon-22786

11/09/2015, 3:32 AM

heh

witty-crayon-22786

11/09/2015, 3:32 AM

@ancient-stone-50795: will comment with a shipit on that one shortly.

ancient-stone-50795

11/09/2015, 3:32 AM

k, thanks.

witty-crayon-22786

11/09/2015, 3:34 AM

@ancient-stone-50795: commented

ancient-stone-50795

11/09/2015, 3:53 AM

@witty-crayon-22786: let me know if https://rbcommons.com/s/twitter/r/3101/diff/2#index_header is what you had in mind

witty-crayon-22786

11/09/2015, 3:57 AM

@ancient-stone-50795: sortof... rather than

self.incremental

, it should be

vts.is_incremental

i think

ancient-stone-50795

11/09/2015, 4:00 AM

yeah, not sure where I got that from

witty-crayon-22786

11/09/2015, 4:02 AM

hopefully it would fail some tests 😃

ancient-stone-50795

11/09/2015, 4:02 AM

Yes, it did!

witty-crayon-22786

11/09/2015, 4:02 AM

hooray

ancient-stone-50795

11/09/2015, 4:03 AM

I thought I had re-run my internal test before publishing but obviously I didn't.

ancient-stone-50795

11/09/2015, 4:03 AM

OK, I just did that time and it passed and I re-pushed.

ancient-stone-50795

11/09/2015, 4:09 AM

So, I don't know what you were thinking about for how to tell zinc to turn off incremental compile for a target, but if there were an interface on VersionedTargetSet that you could ask it to really REALLY invalidate the target, we could add that check to the new

should_compile_incrementally(vts)

and our plugin wouldn't have to reach out and delete the analysis file.

ancient-stone-50795

11/09/2015, 11:47 AM

Effectively, the same as setting

vts._is_incremental=False

, but its not clear to me that right now that just setting _is_incremental=False would turn it off and keep it off, it depends on when

vts.create_results_dir()

is called.

ancient-stone-50795

11/09/2015, 12:08 PM

@enough-analyst-54434: I worked this weekend to get the latest version of pants to work in our repo. I committed https://rbcommons.com/s/twitter/r/3101 to make it into your hotfix release this morning.

enough-analyst-54434

11/09/2015, 2:37 PM

Got it

ancient-stone-50795

11/09/2015, 2:40 PM

Ship it!

enough-analyst-54434

11/09/2015, 3:25 PM

Alright - release is out https://pypi.python.org/pypi/pantsbuild.pants/0.0.57 And that did the trick for Medium

ancient-stone-50795

11/09/2015, 3:27 PM

Thanks, I'm building a new pex for Square

mammoth-pharmacist-54215

11/09/2015, 6:24 PM

thanks everyone for the feedback on https://rbcommons.com/s/twitter/r/3094/ it sounds like people would be much more comfortable with this if the goal was called something like nuke or wash. Everyone okay with me creating a new top level goal that does this called nuke?

happy-kitchen-89482

11/09/2015, 6:25 PM

I’d be comfortable with that

happy-kitchen-89482

11/09/2015, 6:25 PM

I like to err on the side of evocativeness

happy-kitchen-89482

11/09/2015, 6:25 PM

And in fact after that we could rename clean-all to clean, because that “all” is just confusing.

happy-kitchen-89482

11/09/2015, 6:25 PM

We can leave clean-all as an alias for a long time