IRC channel logs

2021-10-25.log

back to list of logs

<zimoun>hi!
<civodul>hey zimoun!
<civodul>did you have more comments on 'guix shell'?
<zimoun>Specifically me. ;-) I have not reviewed or tested more than the 2 first. I assumed that what others said before v3 was enough. I mean, I have nothing special to say about item #1 of v3. And the 2 others items LGTM. I can do a last look now if needed. Tell me. :-)
<zimoun>civodul: so I am doing right now. :-)
<civodul>zimoun: oh you don't have to, i'm just asking because you were one of those who commented
<zimoun>civodul, thanks for asking. :-) I am going to a least apply the serie and test it. :-)
<zimoun>civodul, the environment variable GUIX_ENVIRONMENT is not renamed, right?
<zimoun>civodul, I am going to send to 50960 my general remarks, since you asked me.:-)
<civodul>zimoun: it's not renamed, indeed
<civodul>you don't *have* to send something, i was just wondering whether you had something to add :-)
<zimoun>civodul, now I have. :-) Minor comments.
<civodul>alright!
<zimoun>civodul: done. Thanks for this nice new command. :-)
<zimoun>I agree with Dave’s comment: «Just some things to think about. I just don't want to see Guix committo an interface that slams the door on future improvement due tocompatibility reasons. I mean, what would you even name the nexttool? guix... biome?»
<civodul>heh, who could disagree?
<zimoun>bah if we all agree, why do we add another subcommand instead of improving the interface of the current one? :-)
<civodul>oh, i thought we were long past that discussion
<civodul>see https://issues.guix.gnu.org/38529#17
<civodul>as i see it, we could not find a "reasonable" way to keep the current name and change its semantics
<civodul>plus documents, etc.
<zimoun>I agree about “reasonnable” and I am happy you did it. :-)
<zimoun>However, I am not sure to agree with Konrad’s message you are referring to. Because, if the interface of “guix shell” is not good enough for whatever reason, the next one will be “guix biome”. ;-)
<civodul>yes, and i agree that's a problem
<civodul>i just hope we won't reach that point
<civodul>i'm hopeful :-)
<civodul>but really, it's the first time in 9 years that we have to do that, so my hope may well be grounded
*civodul prepares blog post
<zimoun>For instance, my reply to the mentioned Konrad’s comment <https://issues.guix.gnu.org/38529#21> and <https://issues.guix.gnu.org/38529#25>. Therefore, I am not even sure there is a consensus on the topic. Whatever. Let more forward as your proposal is doing. :-)
<zimoun>civodul: I agree on that Konrad’s comment <https://issues.guix.gnu.org/38529#33> about state a clear backwards compatibility policy though.
<civodul>zimoun: yes, having a policy sounds sensible
<civodul>in the last version of the patch series, i explicitly wrote down a date
<civodul>not quite a policy, but maybe a policy-to-be?
<zimoun>ah, right; May 1st 2023. I have overlooked that.
<zimoun>instead, I would speak about release; until v1.5 for instance. (two releases ahead)
<zimoun>Well, I think this date is a good approximation of the same thing though :-)
<civodul>right, at the current release rate, we don't know when 1.5 will happen, nor whether 1.5 will exist under this name :-)
<civodul>i think it would make more sense to use plain integers like systemd or Firefox
<civodul>or GCC
<civodul>(maybe more a discussion for #guix)
<zimoun>civodul: question asked :-)
<zimoun>civodul, I wrote another drafty blog, for it is worth <https://simon.tournier.info/posts/2021-10-25-software-heritage.html>
<zimoun>in the frame of Disarchive (potential grant) and Open Science prize, I think this kind of stuff is a killer feature; compared to any other tools.
*civodul looks
<civodul>nice
<civodul>i'm late though :-)
<civodul>zimoun: nice high-level view in that post!
<civodul>(typos: "Alice publishes", "Bob redoes")
<civodul>i feel we need to describe the rationale and design of Disarchive somewhere
<civodul>as a followup to https://guix.gnu.org/en/blog/2019/connecting-reproducible-deployment-to-a-long-term-source-code-archive/
<civodul>i've been bugging Timothy about this
<zimoun>civodul: thanks.
<zimoun>Yes, exactly! a follow up. I forgot about this one. :-)
<zimoun>And few people dig til 2019. ;-)
<zimoun>BTW, I had in mind to target Scientists – you know how they often are, if we do not fit *exactly* their world, they lack imagination. ;-)
<civodul>right!
<civodul>what i had in mind is more technical, but we need both
<zimoun>the technical could go to Guix blog and a less technical to HPC one.
<zimoun>do you have news from Timothy?
<civodul>yeah
<civodul>no, no news
<zimoun>argh, for some mysterious reasons URLs like https://git.code.sf.net is rejected by SWH. )-:
<zimoun>And status of https://github.com/scikit-learn/scikit-learn is failed by SWH. Ouch!
<civodul>i guess we have to ask them to investigate?
<zimoun>Already done. :-)
<zimoun>But maybe past 23h is not a good time for asking. ;-)
<civodul>woo
<civodul>right!
<zimoun>hum, I think the ~500 Git URLs failure that Timothy reported should be fixed now. :-)
*zimoun -> zZz