...er. that can't possibly be the error, because pre_execute creates those products if they don't exist.

I owe you a Coke

Attaching an issue with the backtrace / problem details would be great

@witty-crayon-22786: I'm looking at the GroupTask code for the 1st time in a while, but the idea is you should be able to ask for a GroupTask by name from any plugin and get the same group. The concept is certainly dirtied by the flag namespace, but the idea would be to call

group = GroupTask.named('jvm-compile', ['classes_by_target', 'classes_by_source'])

from the contrib plugin.

There are a few issues, including naming (singular/plural), and the product type check not sorting the LHS and RHS when checking for a compatible group

This was done before we had backends though, so it would be cleaner to expose the group in the jvm backend - it would make sense to provide it as a constant I think.

unfortunately i don't have a full trace, because it's occurring in a background thread and python

Then extension backends simply import the constant and use it in their

register_goals

(chained exceptions coming in 3 or so i heard?)

its here in 3 - yes

like so:

python
$ python
Python 3.4.3 (default, Mar 25 2015, 17:13:50) 
[GCC 4.9.2 20150304 (prerelease)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> try:
...   int('2.2')
... except Exception as e:
...   raise Exception('Ooops!') from e
... 
Traceback (most recent call last):
  File "<stdin>", line 2, in <module>
ValueError: invalid literal for int() with base 10: '2.2'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<stdin>", line 4, in <module>
Exception: Ooops!
>>>

There's also a trick to do it in 2

Unfortunately if you're writing polyglot 2/3, I know of no possible way to do it that works in both

I really wish we could write in 3 and run some sort of 3to2 conversion

well - certainly we could write a lib

and the traceback chaining would be our own style, but it would wqork in both

Hm, that's true

Using https://docs.python.org/2/library/traceback.html#traceback.format_exc or somesuch

I've done this in java 2x now for complex chaining where even the bultin does not suffice

basically a tree of exceptions for a tree of execution like we have for workunit logging

I thought there was something in six to help with that

hm

yes, but it does not chain under 2: http://pythonhosted.org/six/#six.raise_from

It does have helpers, but it looks like they mostly just support py3 and don't-break py2

exactly

That's not bad though. I'd love to deploy pants under py3k internally

And we ostensibly support it

getting a py3 CI setup and whittling away errors would be a worthy goal, I started and abandoned in the fall

Oh, btw I've started the ball rolling on Jenkins

Our prod guy is going to look into the boxes we provisioned way back when and get a cluster setup