I (finally) started transitioning our lab’s tools from using parallelism (the multi-splitter) towards collections. In the parallelism scenario, tools were run in their respective working dir, resulting from a chdir performed in extract_parts.sh. That was nice. But I see when using collections that there is a workdir around, but the tool is not actually run in it. I often need to symlink to change input file extensions and can’t imagine noone else does that, so I wondered if I was missing something. If I’m not, then here’s my two questions:
- Is there an accepted place to symlink files in (and how do I pass the working dir in a tool XML)? I don’t want them in the files/ dir, looks bad, they’re not cleaned, etc etc.
- Would it be a bad idea to make tools run inside a workdir, apart from possibly breaking backwards compatibility?
Proteomics systems developer
BILS / Lehtiö lab
Scilifelab Stockholm, Sweden