IRC channel logs

2021-12-03.log

back to list of logs

<zimoun>rekado_: thanks for r-flowclust. The fix is not in c-u-f, right? It is becoming hard for me to know what is broken where. :-)
<rekado_>lots of bioinfo things are still broken on c-u-f
<rekado_>but I'm close to having everything fixed on master
<rekado_>this makes comparing breakage easier
<zimoun>ok
<zimoun>civodul: I am really interested to know more about «“The reproducibility of the study is not possible”» by anonymous reviewers. :-D
<civodul>zimoun: heh, they just didn't do their homework, it's terrible
<civodul>i'm somewhat pissed off by the elitism and arrogance of academic venues
<civodul>but maybe that's just the natural reaction of someone who got their paper rejected :-)
*zimoun feels bitterness from here
<zimoun>Thanks, I discovered https://www.universalrejection.org/ :-)
<civodul>neat, isn't it? :-)
<rekado_>I'm missing context
<rekado_>what happened?
<zimoun>rekado_: https://toot.aquilenet.fr/@civodul/107383353840831620
<rekado_>wow
<rekado_>that's annoying
<rekado_>the comments are priceless
<rekado_>we're applying for funding for reproducibility work and one of our contacts who reviews our submission arrogantly said --- essentially --- that reproducibility with software is a solved problem. You just need to record the versions and metadata! Duh! And something with databases. Gotta have a database.
<rekado_>¯\_(ツ)_/¯
<rekado_>not anonymous, so it's extra surprising to see an opinion like this presented with so much confidence.
<rekado_>we hope that we won't need this person's approval to go ahead.
<civodul>i suppose it means you can add an outreach & dissemination section to your proposal :-)
<zimoun>yeah, I often get the same answer. And I miss then how they answer to question: where does reproducibility crisis come from? More than often, the answer “software is not an issue” is from people doing science, for real, in the 80s. Before the Internet boom. And before the data boom. :-0
<pjotrp>when it comes to rejection we are in good company. Einstein had a lot of trouble too only only got his first appointment *after* he received a Nobel prize
<pjotrp>that was partly racial
<pjotrp>but it is also known that many of the most important papers got rejected or ended up in low ranking journals. Sometime you are ahead of the curve and it hurts
<pjotrp>one reason I helped set up JOSS and BioHackrXiv
<pjotrp>but zimoun is also right: software deployment is only part of the reprocibility crisis. But at least that we can do something about - because we can. And I think it is important.
<pjotrp>reproducibility (I am keyboard challenged)
<rekado_>relevant: https://www.youtube.com/watch?v=DEh1GR0t29k
<rekado_>my colleague wrote about this: in biology it took >20 years for people to recognize crispr, and +30 years to recognize the usefulness of modified RNAs
<rekado_>"the main problem is that the process is completely random, while people's livelihood depends on it"
***zimoun` is now known as zimoun
<zimoun>It remembers this book: https://en.wikipedia.org/wiki/Science_in_Action_(book)
<zimoun>I am not sure to agree with the ontology behind.
<zimoun>But from my memories, it is a good description about the “human“ part of making research
<rekado_>hmm, I'm not offended or anything, but certainly surprised to see that "bless" on c-u-f references slurm, psm, PAM, etc
<rekado_>that's because it's built with openmpi
<rekado_>I feel that this is perhaps a bit excessive, no?
<rekado_>in the conversion of bless from %build-inputs to gexp I also found something odd: the package has both zlib:static and zlib as inputs, but when I specifically request #$(this-package-input "zlib:static") I get the same thing as when I do #$(this-package-input "zlib").
<rekado_>turns out I can build without the static lib, but ... I think that's odd.