nextflow-io / nextflow

A DSL for data-driven computational pipelines
http://nextflow.io
Apache License 2.0
2.7k stars 622 forks source link

Provide a unified start timestamp for scripts and config files #5244

Open bentsherman opened 1 month ago

bentsherman commented 1 month ago

The nf-core pipeline template uses a custom timestamp in the various report file names:

def trace_timestamp = new java.util.Date().format( 'yyyy-MM-dd_HH-mm-ss')
timeline {
    enabled = true
    file    = "${params.outdir}/pipeline_info/execution_timeline_${trace_timestamp}.html"
}
report {
    enabled = true
    file    = "${params.outdir}/pipeline_info/execution_report_${trace_timestamp}.html"
}
// ...

This is problematic because the new config parser does not support top-level variable declarations like this.

The quick solution is to copy the timestamp expression into each report filename. But then there is a small chance that some report filenames might differ by one second since the timestamp is computed multiple times.

The deeper problem I think is that there is not a single start timestamp that can be used across all scripts and configs.

There is workflow.start: https://github.com/nextflow-io/nextflow/blob/12ea4d7c1d9b35d0fc0a8a57836cb33649339b6d/modules/nextflow/src/main/groovy/nextflow/script/WorkflowMetadata.groovy#L256

But the default report file names use a different timestamp: https://github.com/nextflow-io/nextflow/blob/12ea4d7c1d9b35d0fc0a8a57836cb33649339b6d/modules/nextflow/src/main/groovy/nextflow/trace/ReportObserver.groovy#L43

It should be possible to define a single timestamp that can be exposed to config files and also used by workflow.start. It would be another implicit config variable such as launchTime.

We could also try to expose the entire workflow variable, but it isn't resolved until later on, so we would have to do some config parser magic to lazy-evaluate expressions that use the workflow variable.

cc @ewels @pditommaso for your four cents

bentsherman commented 3 days ago

Less important now since a simpler workaround was found:

params.trace_timestamp = new java.util.Date().format( 'yyyy-MM-dd_HH-mm-ss')
timeline {
    enabled = true
    file    = "${params.outdir}/pipeline_info/execution_timeline_${params.trace_timestamp}.html"
}
// ...