<rekado>hmm, even without the package transformation and just an inherited package I get that patch symlink, which breaks the world <rekado>so… instead of creating a package variant of the ‘guix’ package I updated the package definition directly. <rekado>the patch really sped things up considerably. <zimoun`>Can’t wait to see the Jupyter’s demo :-) ***zimoun` is now known as zimoun
<civodul>the demo will be the hands-on session, tho! <zimoun>yeah but I have heard to group 2 could only see and listen <zimoun>rekado: ouch about the world rebuild… Maybe I am saying something silly, is it not possible to grafts to avoid the build rebuild? <rekado>zimoun: it only happens when I build a custom Guix with (package (inherit guix) …) or with options->transfromations <rekado>I think this is a bug that presents itself when the ‘guix’ package is transformed <rekado>since I don’t depend on that transformation, though, it doesn’t block my development of the GWL <rekado>I worked around this by updating the ‘guix’ package in (gnu packages package-management) <rekado>I’ll try to reduce the test case for this and submit a bug report later. <rekado>now I’m back to working on pigx-rnaseq.w <rekado>woo, finish running pigx-rnaseq.w to completion for the first time! <rekado>there are some caching quirks/bugs and a few things about code snippets (e.g. references to non-existent variables don’t trigger an error) <rekado>but: this means that the GWL can in fact be used for non-trivial workflows with arbitrary inputs (the first real-world GWL workflows had hardcoded inputs) <civodul>it's a great idea to try it on a real-world example