https://nf-co.re logo
Join Slack
Powered by
Is the meta naming convention `'meta', 'meta2', 'm...
# linting
p
Is the meta naming convention 
'meta', 'meta2', 'meta3'
, a hard-rule? I'd like to stick to more descriptive names for https://github.com/nf-core/modules/pull/10027.
j
Yes it is to allow consistency and easier mass changes/updates
p
boo-urns
😬 1
j
In the long run we would like to relax it but only if we can have standardised naming
p
I kid. I can see why it would be useful to keep things in check, even if I don't like it like it.
👍 1
j
Yeah aot of people have the same opinions
p
opinions are like blank, everyone's got one?
😬 1
🙂
But cheers for the quick reply, then I can finalise my PR I hope!
🎉 1
j
I may have ulterior motives
p
shared a GIF
m
all will burn soon
😨 1
p
@James Fellows Yates: the linter is not supposed to prevent access to 
meta.<key>
, is it? It throws an error when I try this for example: 
def read_group = meta.read_group ? "-r ${meta.read_group}" : ""
│ bwa/samse          │ modules/nf-core/bwa/samse/main.nf │ main_nf_meta_key: Invalid 'meta' keys detected: meta.read_group, meta.read_group                                                                           │
j
Ah right, yes it is
you must not hardcode meta keys in a module
p
Oh noes
j
As then you start forcing a pipeline developer to comply with the module
And then if you have multuple modules with different key names for the same info - leads to horrible additional work
p
So if I had just merged this PR prior to these new linter rules, this would have been someone else's problem 😄
I can see the rationale though.
j
A reviewer SHOULD have picked up on that
👀
p
Well...this one has been open since February and I'd let it sit there for a while.
So that's on me.
j
Hrm, is suffix allowed now?
p
Nope, that's another error I hadn't mentioned yet.
😄
Are there any examples of how other modules that relied on suffix or metamap.keys were converted? For the latter I'm guessing an input parameter makes sense. But since those can't be made optional (correct?) that'd mean updating all downstream subworkflows/pipelines too, wouldn't it?
(aside from relying on a pipeline specific patch file like the docs mention)
j
I think if you wannted a suffix you needed to put it as a input channel
👍 1
But I've avoided any of this entirely 😉
@Simon Pearce is my arch nemesis on this anyway 😉
p
Oh..so we have to pick sides now? 😶
j
Standards or chaos. Your choice.
😉
m
I have followed the decision on that one, even if I didn't agree with it
p
Thanks for bringing me up to speed btw. I tend to only have time for nf-core intermittently, and don't get a chance to follow along with all the many changes that happen in real-time 😉
j
All good :
🙂
m
Even chaotic agents follow rules sometimes
😅 1
j
Looks at sarek, screams, and runs away
🤣 1
p
<this is fine.meme>?
s
Yeah, I disagree with James and a few other maintainers about this
m
I side with Simon on that one to be honest
5 Views
Open in Slack
PreviousNext
Powered by