IRC channel logs

<civodul>hello!

<civodul>i’m totally swamped with other deadlines but we should make sure to publish the activity report on Friday at the latest

<rekado>oof

<civodul>sounds too ambitious to you?

<rekado>no, it’s a good deadline, but it’s unfortunate that I haven’t been able to finish enough things to add to the report.

<civodul>it’s okay to finish them by Feb. 2025 :-)

<civodul>but yeah, i know that feeling

<rekado>:)

<rekado>I wonder if it would make sense to have at least one short alignment meeting well ahead of the report deadline in the future

<civodul>yes, probably

<civodul>i think i emailed people beginning of December with a clear deadline

<civodul>but that’s not enough

<rekado>the report’s “Perspective” section often bites me a little. It’s what we decide in February should be a good direction, but we don’t really have any structured efforts to arrive there in time.

<civodul>yeah

<civodul>at least not coordinated efforts

<rekado>exactly

<civodul>like you know you’re likely to do stuff around bioinfo packages, i may do stuff around SWH

<civodul>but it’s more like a collection of fuzzy individual goals

<rekado>we might be able to unlock a few more goals or unblock the paths towards those goals if we coordinated efforts just a bit

<rekado>not too much to become yet another responsibility burden, of course

<civodul>yeah

<civodul>worth trying

<rekado>one thing we all can do for this year’s report is to expand each section a little bit to suggest the wider implications

<rekado>e.g. for the guix-bioc section I’d like to emphasize that this means that Guix is now one of the largest distributors of (largely) reproducibly built R code, and with the backing of SWH it’s a solid foundation for long-term reproducibility

<rekado>I feel we sometimes need to do a little bit more to frame our results and sort them into the larger framework of our goals.

<civodul>right, i agree

<civodul>but…

<civodul>“we all can do” -> “we” is going to be the two of us i’m afraid

<civodul>because it’s probably too late and i have to admit i’m tired of trying to coordinate

<rekado>yes, I understand

<rekado>I pushed a few more of my changes

<civodul>neat, thanks!

<civodul>ACTION -> lunch

<rekado>hmm, no substitutes for python-tensorflow from guix.bordeaux.inria.fr.

<rekado>the tensorflow logs are almost useless because they are so large

<rekado>49M!

<rekado>In procedure copy-file: No space left on device

<rekado>note: build failure may have been caused by lack of free disk space

<civodul>ouch

<civodul>that’s C++, right?

<civodul>it would make sense to build with -g0 to save space

<PurpleSym>rekado: Batching work is some motivation to get things done ;)

<rekado>I’ve been trying to build tensorflow for days now. My system keeps killing cc1plus because it exhausts my laptop’s resources.

<rekado>this is fascinating: https://www.phoronix.com/review/radeon-cuda-zluda/

<rekado>it’s no longer actively developed by AMD and targets ROCm 5, but it’s a working drop-in replacement for CUDA on Radeon GPUs.

<rekado>worth packaging

<civodul>HIP is pretty much a drop-in replacement too, no?

<civodul>except one has to s/cuda/hip/ in the source

<civodul>interesting stuff

IRC channel logs

2024-02-12.log