I am seeing this new error in <#CGFUX04HZ> with nf...
# nf-schema
m
I am seeing this new error in #CGFUX04HZ with nf-schema 2.7.1 and NXF latest everything, but not 25.10
Copy code
cf_ploidy (2): Value is [integer] but should be [string]
n
We had to implement some kind of type casting for parameters in strict syntax that doesn't use typing. It could be caused by that. Might be good to allow both types for now. I'll try and see if I can decrease the amount of false negative validations sometime later
m
ok
This is going to break some things
I'll try to make sure all works
n
Allright! Please send me more issues if you find them
m
As usual
keanu thanks 1
n
Could you send me the exact command you used that caused that issue? I'm trying to find some workarounds for smarter type casting until static typing is officially in Nextflow
m
I see it in the CI for many tests
-profile test --aligner bwa-mem --save_reference skip QC/recal/md
is the first one in the first shard
so
nf-test test ... tests tests/aligner-bwa-mem.nf.test
should trigger it
n
Okay thanks! This confirms my suspicion, I'm working on a fix
m
That was fast
🚄 1
n
All these issues will be solved once static typing is introduced so currently it's all a bit hacky while we're waiting for that feature
Your specific issue here is because
cf_ploidy
is a String value consisting of an Integer: https://github.com/nf-core/sarek/blob/4bd2948f98c5bf7b785c91cf6708fffccab25467/nextflow.config#L77
The real issue we needed to use type casting for is when user specify parameters via the command line because those are always String values in strict syntax. I'm working on limiting the parameters that get converted to the ones that are specified on the command line instead of all String parameters. It's not perfect, but should fix a lot of issues
m
Nice ❤️
Trying it out
🤞 1
LGTM
thanks
n
Perfect!
Setting the parameter via the CLI like this
--cf_ploidy 5
will still fail but that was also the case with the v1 syntax parser
It's always a good idea to allow both
string
and
integer
for parameters that could also possibly be an integer
party this 1