Mash info buffering about poppunk

[johnlees]

I'll look into improving the interface with the mash output, as I think this also relates to issue #9 taking up far more memory than necessary

[nickjcroucher]

No problem at all on this one with Mass, Southampton or Nijmegen datasets, but falls over pretty much immediately on Maela:

Traceback (most recent call last):
File "./PopPUNK/poppunk-runner.py", line 9, in
main()
File "/lustre/scratch118/infgen/team81/nc3/strain_structure/comparative_clustering/PopPUNK/PopPUNK/main.py", line 230, in main
sketch_sizes = getSketchSize(ref_db, kmers, args.mash)
File "/lustre/scratch118/infgen/team81/nc3/strain_structure/comparative_clustering/PopPUNK/PopPUNK/mash.py", line 173, in getSketchSize
mash_info.wait(timeout=15)
File "/lustre/scratch118/infgen/team81/nc3/nc3/lib/python3.6/subprocess.py", line 1449, in wait
raise TimeoutExpired(self.args, timeout)
subprocess.TimeoutExpired: Command 'mash info -t ././maela_db/./maela_db.14.msh' timed out after 15 seconds

I need to have a think about this. Reopening as a reminder.

[johnlees]

This is because the mash info command takes longer than 15s to run?
Perhaps the best thing to do is no longer poll for successful command completion, and instead check whether sketch was set, and also raise RuntimeError if more than two lines are read

[johnlees]

I think just removing lines 173-175 of mash.py would work

[johnlees]

Does that commit work?

[nickjcroucher]

I think that would make sense, we’ve already got the except for error catching.

[nickjcroucher]

That seems to do the trick, worked fine.