Conversation
The pretext/pretext2asm, pretext/juicerc, and pretext/makepairs modules explicitly throw errors when running with Conda. Add them to the conda exclusion list in skip_nf_test.json to prevent CI failures in conda runs. Also add subworkflows/sanger-tol/zippypretext which depends on these pretext modules and would also fail with Conda.
|
@copilot resolve the merge conflicts in this pull request |
# Conflicts: # subworkflows/sanger-tol/busco_gene/main.nf
|
@copilot resolve the merge conflicts in this pull request |
Resolved. There was one conflict in |
The merge conflicts were already resolved in commit |
- Fix container_links: change docker.io/ubuntu:20.04 to quay.io/biocontainers/ubuntu:20.04 - Fix meta_name: update meta.yml name from 'makepairs' to 'pretext_makepairs' to match process name PRETEXT_MAKEPAIRS
- Fix container_links: change ghcr.io/sanger-tol/agp-tpf-utils:1.3.3
to quay.io/sanger-tol/agp-tpf-utils:1.3.3 (quay.io is an allowed prefix)
- Fix correct_meta_outputs: update correctedagp output pattern in meta.yml
from '*_corrected*.agp' to '${meta.id}_corrected*.agp' to match main.nf
DLBPointon
left a comment
There was a problem hiding this comment.
More comments, most are simple things
Co-authored-by: Damon-Lee Pointon <51855558+DLBPointon@users.noreply.github.com>
Co-authored-by: Damon-Lee Pointon <51855558+DLBPointon@users.noreply.github.com>
Co-authored-by: Damon-Lee Pointon <51855558+DLBPointon@users.noreply.github.com>
Co-authored-by: Damon-Lee Pointon <51855558+DLBPointon@users.noreply.github.com>
Co-authored-by: Damon-Lee Pointon <51855558+DLBPointon@users.noreply.github.com>
BUSCO v6.1.0 performs a mandatory online version-check at startup (downloading file_versions.tsv from busco-data.s3.amazonaws.com and its backup at busco-data2.ezlab.org) even when --download_base_url is explicitly set. When both servers are temporarily unreachable from CI runners, this causes the busco_gene subworkflow tests to fail after ~10 minutes of retries. Fix: - Add a GitHub Actions cache for the bacteria_odb10 lineage tied to the BUSCO container digest. A pre-download step populates the cache on first use; if the BUSCO server is down the step is non-fatal (|| true) and the cache is simply left empty for this run. - Export BUSCO_LINEAGES_PATH only when a versioned lineage directory (bacteria_odb10_YYYY-MM-DD) is confirmed to exist. - Read BUSCO_LINEAGES_PATH into a busco_lineages_path Nextflow param in the shared nf-test.config. - In the busco_gene test config, make ext.args a closure: use --offline (skipping the online check) when a cached lineage path is set, otherwise fall back to the original --download_base_url. - Update both busco_gene tests to pass the cached lineage directory as input[2] when available.
There was a problem hiding this comment.
Copilot has added this to the actions again
There was a problem hiding this comment.
Can you add ontologies so it's in line with the other modules:
- fasta:
type: file
description: Assembly summary file
pattern: "*.(fasta|fa|fna)$"
ontologies:
- edam: "http://edamontology.org/format_1929" # FASTA
| ]) | ||
| input[1] = channel.of([[ id:'test' ], 'bacteria_odb10']) | ||
| input[2] = [] | ||
| input[2] = params.busco_lineages_path ? file(params.busco_lineages_path) : [] |
There was a problem hiding this comment.
This shouldn't be needed. There isn't any ambiguity with your input, it's []
| ]) | ||
| input[1] = channel.of([[ id:'test' ], 'bacteria_odb10']) | ||
| input[2] = [] | ||
| input[2] = params.busco_lineages_path ? file(params.busco_lineages_path) : [] |
There was a problem hiding this comment.
This shouldn't be needed. There isn't any ambiguity with your input, it's []
| // When set (e.g. via BUSCO_LINEAGES_PATH env var in CI), BUSCO tests use this | ||
| // pre-cached download directory and run with --offline to avoid network failures. | ||
| busco_lineages_path = System.getenv("BUSCO_LINEAGES_PATH") ?: null |
There was a problem hiding this comment.
Copilot again adding stuff where it shouldn't be. This isn't needed it isn't an ambiguous field when we are testing.
There was a problem hiding this comment.
not going to use copliot that often in the future ...
DLBPointon
left a comment
There was a problem hiding this comment.
Most comments, mostly around Copilots additions again.
The GitHub Actions job "x64 | conda | 1" (job ID 80538385830) was failing because the
pretext/pretext2asm,pretext/juicerc, andpretext/makepairsmodules all explicitly throw errors when run with Conda:These modules were missing from the
.github/skip_nf_test.jsonconda exclusion list, so when their tests ran in the conda CI job, they failed immediately. TheJUICERCtest was also failing because it usesPRETEXT2ASMin its setup block.Changes Made
.github/skip_nf_test.json: Added"modules/sanger-tol/pretext"and"subworkflows/sanger-tol/zippypretext"to thecondaskip list so these tests are excluded from conda CI runs, consistent with how other no-conda modules (e.g.longranger,ancestral,telomere) are handled.