use_relative_output_paths

[mlin]

#604 #606

@rhpvorderman Here's a refactor based on my earlier comments, WDYT? (Opened a new PR because I didn't have permission to push to your branch, but keeping the commit history will ensure your co-authorship when we merge)

[coveralls]

Pull Request Test Coverage Report for Build 3384144984

• 35 of 36 (97.22%) changed or added relevant lines in 1 file are covered.
• 4 unchanged lines in 3 files lost coverage.
• Overall coverage decreased (-0.05%) to 95.161%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
WDL/runtime/task.py	35	36	97.22%

Files with Coverage Reduction	New Missed Lines	%
WDL/runtime/download.py	1	88.8%
WDL/runtime/workflow.py	1	97.17%
WDL/runtime/task.py	2	95.5%

Totals
Change from base Build 3376681751:	-0.05%
Covered Lines:	7197
Relevant Lines:	7563

---

💛 - Coveralls

[rhpvorderman]

I did a test. Unfortunately, when using the call_cache it outputs everything in the same directory, without keeping the directory structure. This was not the case before the refactor.
I am checking the code to see if this can be fixed easily.

[rhpvorderman]

A case is missing for call cached entries. These also have an outdir. I wonder if it is possible to "know" that the code is taking a path from the cache.

[mlin]

@rhpvorderman Good catch, I'll look into this

[mlin]

@rhpvorderman updated, please give this a try? I think the logic is now very similar to your original, just with a few added heuristics to make it harder (but not impossible) to confuse it with other subdirectories named work/ or out/.

[rhpvorderman]

@mlin, this works great! I tested it with call caching and output hardlinks, and both seem to work fine. Thanks!

[rhpvorderman]

@mlin, thanks again for your efforts!