I would like to roll out linting to all pipelines ...
# infrastructure
p
I would like to roll out linting to all pipelines / modules with the non-failing-but-PR-comment approach that I wrote for the Nextflow training (example). What's the easiest way to do this? Can we centralise it somehow and apply it everywhere without too many multi-repo code changes? πŸ‘€
Happy to ditch the formatting part of it (to start with / maybe forever)
This is mostly a transition infra for just this year. Once we have adopted
nextflow lint
and auto formatting it can go into pre-commit / whatever and be part of the regular CI
But the comments are nice as an early thing so that we don't make it blocking
m
happy to have this in the pipelines, but can we ignore all code from nf-core/modules as the specs for these can vary compared to what we might have in pipeline?
p
Specs how? I wondered about ignoring modules, but I figured that it doesn't hurt as it'll hopefully prompt people to upgrade / fix modules
And Nextflow doesn't care where the error is, it'll still refuse to run
m
yeah I know, but I'm just concerned that it might confuse people into where they should fix the issue
πŸ‘πŸ» 1
ie in the pipeline or upstream in the modules repo, or even worse in a plugin
and by specs I meant more or less how we refactor syntax and code from time to time without actually changing any actual behavior
more thinking about the warnings here than the actual linting errors
s
Don't think auto formatting copes with Harshil alignment still, so I'm not a fan of imposing auto formating
party this 1
πŸ’― 2
We should fix all the errors in nf-core/modules at some point πŸ˜‰
πŸ˜† 1
p
Yeah auto-formatting also still deletes comments
Once that's sorted I think we should turn it on though
I don't really care about Harshil alignment inconsistencies, I think we can just update our docs if needed homer disappear
shake no 1
shakefist 1
meow no 2
meow tableflip 3
πŸ˜† 1
The beauty of auto-formatting is that no-one ever needs to think about it again
s
But it looks ugly
party this 3
r
i still need to reaqd it though. or claude is getting free reign πŸ˜„
p
Is it really that much worse? eyebrows
Coming from a standpoint of having spoken to Ben about it and him saying that the formatter might need a complete rewrite to be able to handle it πŸ˜…
My point is that it'd be nice to solve these small things, but I'm not sure that the small changes in Harshil formatting are enough of a reason to not use the feature entirely.
If we can iron them out then great, but I don't see it as a reason to not go ahead
(vs. deleting comments, which I do)
m
If you remove Harshil alignment I quit nf-core 🀣
🀣 3
πŸ™€ 2
s
But you happily use the auto formatter?
m
Because I keep hoping it will restore the alignment
I praise consistency, but I'm #C0A2RDVD9L7 forever
p
Ok maybe I need a refresher on what's wrong and why everyone is so upset about it πŸ˜†
m
I think it's just that it makes the code look good, and sentimental reasons. We all care about this formatting. And it'll be sad to see it go
s
oh there is a lot of sentimental attachment, it isn't very rational πŸ˜‰
πŸ˜… 1
meow love 1
πŸ’― 1
p
shared a GIF
m
shared a GIF
j
this is a github action right? I think we would have to add it to the template. unless we run it periodically πŸ€”
p
Yeah I was kind of hoping someone would have an idea of how to do it without touching the template πŸ˜…
The pipelines who will need it the most are probably also the ones that are slow to adopt template merges!
πŸ™Š 1
🀣 1
j
we could run it periodically for all pipelines no? and maybe use issues instead of PR comments
but I am not sure how easy would this be