IRC channel logs

2022-01-14.log

back to list of logs

<zimoun>hi!
<civodul>howdy zimoun!
<civodul>are we publishing Arun's post today?
<zimoun>civodul: pushed half hour ago, I guess. :-)
<zimoun>for the interested reader: https://op.europa.eu/en/publication-detail/-/publication/6bc538ad-344f-11eb-b27b-01aa75ed71a1/language-en
<zimoun>it is about Reproducible & co.
<zimoun>I have not read all yet. However, I find really weird that almost all people around this topic speak about “Open”. But Open is not the key for reproducing. The key is twofold: lookup and capture all the computational stack. Otherwise, as much as Open it is, it is doomed. I have plenty of Open examples impossible to reproduce.
<civodul>zimoun: yay!
<civodul>thanks!
<civodul>zimoun: this report looks somewhat empty, no?
<civodul>the recommendations look rather vague and consensual to me
<civodul>(consensual is not a bad thing, but i mean there's nothing new here)
<zimoun>Empty, I do not know. Some sentence are weird: «As reproducibility is a relatively new topic» What?!? It is as old as Galileo :-)
<zimoun>civodul: my point is just that in “Open data” section, it is consensual, sure, but in the same time, it provides a message as if “Open“, share, etc, will fix the issue. It is just wrong and a clear misunderstanding of the issue. IMHO.
<zimoun>and it is an European Union publication. Potentially driving funds.
<civodul> https://hpc.guix.info/blog/2022/01/ccwl-for-concise-and-painless-cwl-workflows/ yay!
<civodul>zimoun: yes, so it's a bit of a concern that it seems to not quite show the current state of reflections
<civodul>that they shamelessly reuse those (nice) visuals from others also doesn't make it sound serious
<civodul>these are great illustrations, but it contributes to the impressions that the authors didn't look very deeply in those matters
<civodul>though to be fair there are also many references
<civodul>dunno
<zimoun>hehe! Yeah!! Somehow, it is a good Friday troll. :-)
<zimoun>rekado_: about the recent ccwl post, I am thinking if it is possible to somehow type the inputs/ouputs for GWL? Instead of basically an unique string type.
<civodul>also: https://www.softwareheritage.org/2022/01/13/preserving-source-code-archive-files/ !
<zimoun>I dropped an email to Timothy 3 days ago for a Guix blog post about this topic. :-)
<zimoun>That’s really cool!
<zimoun>civodul: «On the one hand, advocating the adoption of SWHID by packaging tools, that will take quite some time» Ahah!
<civodul>heh
<civodul>i'll end up writing that blog post at some point
<civodul>zimoun: looks like i won't visit Académie des Sciences: https://osec2022.eu/ :-/
<zimoun>civodul: oh no!
<zimoun>So disappointed… I was so excited to visit Académie de Sciences.