file() behavior changed with glob on s3 with latest edge release compared to latest stable release

[maxulysse]

Bug report

Expected behavior and actual behavior

file() used to work with globs on s3 buckets, but no longer with the latest edge release.

Issue was noticed while trying to reproduce issues with the nf-core/proteinfold pipeline.

Steps to reproduce the problem

workflow {

ch_input = file("s3://ngi-igenomes/test-data/proteinfold/db/alphafold_mini/small_bfd/*")

println ch_input

}

Program output

export NXF_VER=23.04.2; nextflow run .
N E X T F L O W  ~  version 23.04.2
Launching `./main.nf` [magical_planck] DSL2 - revision: b71c613cc3
[/ngi-igenomes/test-data/proteinfold/db/alphafold_mini/small_bfd/bfd-first_non_consensus_sequences.fasta]

export NXF_VER=23.06.0-edge; nextflow run .
N E X T F L O W  ~  version 23.06.0-edge
Launching `./main.nf` [lonely_wozniak] DSL2 - revision: b71c613cc3
[]

Environment

• Nextflow version: [?]
• Java version: [?]
• Operating system: [macOS, Linux, etc]
• Bash version: (use the command $SHELL --version)

Additional context

I think we should be using Channel.fromPath() in that case, but wondering if this change of behavior of file() is expected or not.

[drpatelh]

We have just observed the same behaviour again with nf-core/viralrecon as reported in nf-core/viralrecon#385

Works with 23.04.2 and fails with 23.07.0-edge. Be good to know whether this change was intentional or is a bug because it will mean current versions of pipelines will break if a backward-compatible fix isn't pushed to NF.

[pditommaso]

Too bad, looking into it

[pditommaso]

Solve via 19a72c4. Thanks for reporting it