Hey, when talking with <@U049VM24VJR> about <#C08M...
# nf-test
n
Hey, when talking with @Johannes Kersting about #C08M024RZ5K we saw that in his
<http://default.nf|default.nf>.test.snap
something weird occurs here. This section is produced by snapshotting the
stable_path
, but it shows md5 hashes of
<http://main.nf|main.nf>
files, which are definitely not part of the pipeline output. When looking a bit closer we can see that this big list structure is located where one would expect the plain string
modules
(because his pipeline produces a directory 'modules' that contains the discovered disease modules). Now with a bit of agentic help I found out that nf-test checks for all strings it receives if they exist as a file/directory in the working directory from which nf-test was launched. So if nf-test is launched from the pipeline root directory, the string
modules
is interpreted as a reference to the
modules
directory in the pipeline, and thus all the files in the directory are hashed. Now the easy workaround would be to rename the
modules
subdirectory in the pipeline outdir to something that is not present in the pipeline. But I think this is something that could use a more structural fix. However, I am not sure if this is something that should be tackled by changing how nf-core calls nf-test, or by altering nf-test.
😅 1
neon cat spin 2
Also tagging @Judith Bernett as you were interested
👀 1
Tagging @mashehu as I don't know who is the best person to have a look at this
m
confused dog
m
cc @Nicolas Vannieuwkerke because it sounds like an nf-test bug
m
do you have this when you run your tests locally, or is it on CI as well?
n
It seems to happen both in the CI and locally, as the snapshots match and the tests pass
n
Wow this sounds like quite a nice bug, thanks for reporting! Would you mind opening an issue in nf-test for this?
n
Will do!
🙌 1
Created via #365
n
Thank you!!!!