IRC channel logs


back to list of logs

***rekado_ is now known as rekado
<zimoun>rekado_: how can I use GWL with “./pre-inst-env”? Using GWL using installing it?
<rekado_>the GWL’s pre-inst-env or Guix with pre-inst-env?
<rekado_>guess I’ll have to update the GWL’s pre-inst-env script now
<zimoun>Ok, that’s what I did but I thought there was a “another” way.
<zimoun>So, the path in GUIX_EXTENSIONS_PATH should be fixed and documented. See
<civodul>zimoun: are you applying Pjotr's patch (edited) to the repo?
<zimoun>civodul: done
<civodul>thanks, zimoun!
<civodul>so i guess we'll publish tomorrow if that's fine with y'all
<zimoun>some TODO should need to be removed. And maybe Perspective a bit extended.
<civodul>yeah, i'll take a look tomorrow morning i think
<civodul>roelj: did you check the "Personnel" part?
<roelj>civodul: No, let me check
<roelj>Yeah so, I don't have a fixed time-per-year, so it's probably more like 0.1 person-year.
<zimoun>roelj: you were at 1 person-year and now it is 0.1 person-year, correct?
<roelj>That's the effective change in the reports, yes.
<rekado_>zimoun: about GUIX_EXTENSIONS_PATH: I really don’t know how to move forward. My feeling is that we shouldn’t add .config/guix/current because that location is arbitrary.
<zimoun>rekado_: do you have something more or new to say about the bimbs channel? For the annual report. Just to be sure we are not missing the obvious. :-)
<rekado_>not sure if that’s report-worthy
<rekado_>we’ve been shifting packages to guix-past and will be shifting some to guix-science
<rekado_>the long term goal is to reduce the size of guix-bimsb and use it only for the few things that have no place in guix-past and guix-science
<rekado_>I don’t think it’s worth mentioning this in the report
<zimoun>and bimbs-nonfree?
<rekado_>shouldn’t be mentioned at all ;)
<rekado_>our nonfree channel also contains an unfortunate set of free packages that have nonfree dependencies.
<rekado_>there are a few of those on bioconductor
<zimoun>Maybe it is worth to add a sentence to say that generic packages are moved to Guix Science and other to Guix past. Since it appears last in the section. Remaining specific packages in the bimbs channel. I do not know.
<rekado_>(when people compare free software requirements as being a “religious” devotion this is an example of a real inconvenience of not separating free stuff from nonfree stuff.)
<zimoun>I agree.
<zimoun>rekado_: again about remaining TODO, did you training that you would like to mention? Or let remove this TODO in the «Training Sessions».
<rekado_>no trainings to report
<rekado_>we wanted to do online trainings but the platform here isn’t ready
<zimoun>ok. thanks.
<zimoun>btw, keep me in touch if it is not in German. ;-)
<rekado_>they will be in English, but I don’t think they will be public :-/
<zimoun>civodul: I pushed a tiny update. Ah sorry, I have not double-checked the commit message I pushed on the behalf.
***zimoun` is now known as zimoun
<civodul>zimoun: thanks for your work on the article!
<civodul>i'm proof-reading the SWH part
<civodul>(removing a few links since that looks bad in the PDF output otherwise :-))
<zimoun>cool, I have not tried to build the PDF. :-)
<civodul>guix build -f ../doc/build.scm
<zimoun>ah good to know :-) I am only using haunt.
<zimoun>BTW, adding ’–hard-dereference’ in (guix docker) seems to fix my issue about “guix pack -f docker”. Even if I have no clue why it happens. Another story.
<civodul>oh right, i have a diff for that
<civodul>and an explanation
<civodul>well, what i already wrote i gues
<zimoun>I understand the tar part, but I do not understand why on one machine it does link and the other not. I have checked the daemon option, gc a lot, rebuild in case I miss something, still difference. I have to live with it I guess. :-)
<rekado_>zimoun: the GWL now uses the same Guix that the time-machine spawned:
<civodul>zimoun: did you use offloading on one of these machines?
<zimoun>civodul: no, as far I remember.
<zimoun>rekado_: cool!
<zimoun>rekado_: I do not know if you watched Pierre’s talk about REPL. He wrote a post and it is interesting, another view about graph instead of pipeline. In case you need some reading to asleep ;-)
<rekado_>I haven’t watched the talk (I only had time for two) but I skimmed the blog post.
<rekado_>to me enthusiastic shell replacements are always full of ideas, but I feel I probably wouldn’t use them
<rekado_>the closest I get to a practical shell replacement in my workflow is eshell with custom commands
<rekado_>everything beyond that I feel is better done as a Guile program
<civodul>i kinda feel the same
<zimoun>Me too. I have tried different variants. And basically, the best for me is Emacs with custom stuff otherwise “M-x shell“ with plain Bash.
<civodul>maybe we're old farts, rekado_
<zimoun>Well, I find interesting to see the REPL as working with a graph. And Pierre claims with SLY (which is better than Geiser, troll! ;-)) you can compose the graph interactively. I do not know.
<rekado_>I didn’t feel truly old just a few months ago, but feedback from my environment has provided enough evidence that I am, in fact, an old fart.
<rekado_>doesn’t mean that the shell is great. It just *is* — not just here but everywhere.
<rekado_>that is its primary value to me.
<rekado_>(a bit like the primary value of ‘vi’ being ubiquity)
<zimoun>I agree, I had long discussion about that with Pierre. :-)
<rekado_>I welcome explorations like that, though. I *want* to see different ways to interact with computers.
<bavier[m]>toggle switches... :)
<rekado_>I also want to see a revival of the old-fashioned lisp machine UI
<rekado_>if they have a reassuring click and are a little heavy: I’m all for it!
<zimoun>lisp machine? Do you mean Guix+Emacs? ;-)
<rekado_>Guix + Emacs is all lispy, but it doesn’t give me full introspection and run-time modification abilities
<rekado_>within Emacs you can describe anything and re-eval a definition to change the behavior
<rekado_>I want Emacs for *everything*, for graphical widgets, etc
<rekado_>I want values to retain their type upon processing: files whose name I print should retain a reference to the file (that’s one of the things I was surprised to see in Pierre’s blog post)
<zimoun>Pierre is exploring such ideas, especially with Nyxt, IIUC.
<civodul>we all dream about the Lisp machine UI... but was it that good?
***rekado_ is now known as rekado