Cool - thanks @mvdbeek - I'll work on getting a release out with this in it.

Shouldn't that be fixed upstream in cwltool?

@nsoranzo I imagine it was but we have a fixed older version of cwltool as a dependency, I'll remove this line when I merge my cwltool update as part of #714. Does that sound like the right approach?

xref common-workflow-language/cwltool#450

Perfect, thanks @jmchilton!

@nsoranzo Cool - thanks! I've made that change in #714 and rebased it as d2ea830.

I'm going to see what the tests say about #723 and if things look okay I think I will do a release so this is fixed and Conda is fixed as well. Then I will merge #714 later this week (or maybe later today) and create a new set of problems.

Well so things went sideways because the new galaxy-lib introduced test errors (these can be seen in #723 but would be experienced in the wild anyway since we don't pin an upper bound on galaxy-lib). So I just cleaned up and merged my CWL WIP with 2462099 and did a release so either the dependencies are fixed when coming from PyPI ... hopefully ... but probably not and we shall see.

@jmchilton I couldn't see any new test error in https://travis-ci.org/galaxyproject/planemo/jobs/272496388:

• jobloaderctx is caused by using an old cwltool, is fixed by commit 2462099
• fleeqtk not found: I'm sure I've seen this in previous Travis builds
• AttributeError: 'module' object has no attribute 'numproc': should be fixed by [16.01] Backport uwsgi fix from #2836. galaxy#4565

Am I missing something?

I just hadn't seen the jobloaderctx error previously - my 👀 aren't 🦅 (they call me 💩👀). I didn't think it would have occurred with older galaxy-libs before the two I released today.

jobloaderctx error is a side-effect of pinning ruamel.yaml in this very PR, in fact it started with https://travis-ci.org/galaxyproject/planemo/jobs/272481174