Harden cut command in Snakefiles

[chillenzer]

Found this bug with Arthur. Hemera just hit the next digit in job ids and these cut commands were not up to that challenge. The one for LSF is an educated guess from what the internet says the output should look like. I think I've never run on an LSF cluster.

[ikbuibui]

Moving from characters to fields looks okay, but without looking at the input file I cannot say anything about correctness. If this has been tested as working, its okay for me

[JessicaTiebel]

Thanks for fixing this bug, during my absence! Arthur already told me that it works and it looks good to me. For the LSF case I just don't understand why you used the cut command with other settings. You probably have a reason, if you let me know I will approve this PR. 😃

[JessicaTiebel]

Arthur told me, that the cut -d' ' -f4 command worked on hemera. Why did you choose cut -d'<' -f2 here? What is the difference?

[chillenzer]

The -d sets the delimiter (to whitespace resp. </> in this case) and the -f selects the "field", i.e., the colum with respect to this delimiter (indexing starts with 1). So, cut -d' ' -f4 means: Split the string with respect to whitespaces and select the fourth field resulting from that.

The LSF version: My best guess was that the output from bsub is as shown in this stackoverflow answer. The two cuts split the string into

Full: "Job <930> is submitted to default queue <normal>."
        1  ^          2                        ^   3
First cut: "XXX> ..."
           1   ^   2
Second cut: "XXX"

Could have been a regex but this was faster.

[JessicaTiebel]

Thank you! This seems very reasonable.

[chillenzer]

The slurm version has been tested on hemera. LSF is untested.

[JessicaTiebel]

I wrote the LSF version for the Summit cluster. As far as I know, Summit is now decommissioned and I don't know if any other relevant cluster is using LSF. So testing is difficult.