IRC channel logs

2022-02-01.log

back to list of logs

<rekado_>it’s from 2019 but it references our blog post: https://www.cyphar.com/blog/post/20190121-ociv2-images-i-tar
<civodul>ah yes, but they were misguided by the title it seems
<rekado_>yes
<rekado_>but I found it to be full of information on tar that I didn’t know I didn’t want to know :)
<civodul>heheh
<civodul>zimoun, do you copy me?
<civodul>i think we now have most constributions at https://gitlab.inria.fr/guix-hpc/website/-/blob/master/drafts/activity-report-2021.md
<civodul>i'd welcome comments/suggestions
<civodul>and... if someone is inspired for the closing, don't hesitate :-)
*rekado_ looks
<rekado_>a verb is missing here: “On one hand, GNU Guix allows us to ensure an experimental software environment [is?] reproducible across various high-performance testbeds.”
<rekado_>“we can […] describe the experimental environment as well as the experiences[?] themselves” — really “experiences”? Or should this actually be “experiments”?
<rekado_>“publishing documents is automatized using continuous integration” – I think “automatized” should be “automated”.
<civodul>"experiments" yeah (a false friend in French)
<civodul>noted
<rekado_>“tranform” –> “transform”
<rekado_>“monolitic” -> “monolithic”
<rekado_>“reproductible” -> “reproducible”
<civodul>the next section has a few typos too
<rekado_>“took the opportunity of the writing of an article” —> “took the opportunity of writing an article”
<rekado_>should I skip it and come back to that section later?
<civodul>yes, you can skip that one
<rekado_>ok
<rekado_>I’ll make a minor change to the GWL section. Has some clunky wording
<civodul>ok!
<civodul>should we merge the two workflow sections?
<rekado_>yes
<rekado_>they are both unusually short
<rekado_>I just noticed that technically the release of GWL 0.4.0 would have to go into the next report … :)
<rekado_>happened in 2022!
<rekado_>“Achille’s heel” –> “Achilles’s heel”
<rekado_>or “Achilles heel”
<rekado_>anyway, the guy’s name is “Achilles”
<rekado_>wikipedia says it’s “Achilles' heel”
<rekado_>punctuation is my Achilles heel.
<rekado_>“and this is a significant difference compared to other buildpacks” — I get why this interjection exists, but it doesn’t actually fit, because the interjection is generic while the surrounding sentence is specific
<rekado_>obviousy, other buildpacks don’t use “guix time-machine” and “channels.scm”!
<rekado_>i think this sentence would benefit from mentioning the generic solution first before speaking of the specific (namely “guix time-machine” and “channels.scm”)
*rekado_ commented on the merge discussion on github for good measure
<rekado_>“Guix lets users re-deploy software environment” –> “environment[+s]” or “[any|a] software environment”.
<rekado_>“Since 2019[-,]”
<rekado_>“make that a reality” is vague. The previous sentence only says “possible if source code is available”
<rekado_>this could be changed to “is [+permanently] available”
<rekado_>alternatively, the goal of building a reliable archive of past source code bundles should be explicitly stated before the sentence mentioning the collaboration.
<rekado_>“A lot has been achieved [+since then]”
<rekado_>“Guix expects to be able to download those tarballs and to verify that it matches the expected cryptographic hash” — “it” is a number mismatch. –> “that they match”
<rekado_>“deal with [+this] impedance mismatch”
<rekado_>“This year we deployed on the Guix build farm infrastructure …” – this sentence has a problem. Deployed what? Or is “deploy” the wrong verb?
<civodul>rekado_: it's from Feb. to Feb., so we can have the GWL release here :-)
<rekado_>phew!
<civodul>right in time!
<civodul>let me fix these typos
<rekado_>“However, there is in HPC circles the entrenched perception” — I’d reorder words a little, because I can only read “there is in HPC circles” haltingly. I’d do “However, in HPC circles there is the entrenched perception”
<rekado_>“AVX-512 on x86_64, NEON on ARMv8” – should this be “such as AVX-512 on x86_64, or NEON on ARMv8”? Or is this really a fixed tuple?
<rekado_>“We showed that these concerns are largely unfounded…” — I’d put the articles in chronological order. I was surprised to see the 2018 article mentioned after the 2019 article.
<civodul>"we deployed, on the Guix build farm, infrastructure to continuously ..."
<rekado_>“The latter article showed” <— this would need changing then, too
<rekado_>yes, sounds better
<rekado_>“There remain cases, though, where this technique is not applied.” — sounds like it *could* be applied but is not (for whatever reason). Maybe use “applicable” instead?
<civodul>yes
<rekado_>“A notorious example of packages that do not support FMV are C++ header-only libraries, such as the Eigen linear algebra library.” — this is another numbers confusion. “A notorious example”(singular) “are libraries” (plural), “such as this library” (singular)
<civodul>oops indeed
<rekado_>it’s probably correct, but my head spins when I switch from singular to plural and back again
<rekado_>is it “tailored for” or “tailored to”?
<civodul>well, "an example [...] is"
<civodul>i like that i can learn how to improve on all this :-)
<rekado_>“Guix proposed a dozen of Julia packages” — I don’t know what “proposed” is supposed to mean in this context.
<rekado_>(and I think the “of” can be left off; with “of” it sounds to me like what follows is a commodity with a well-known unit)
<rekado_>“the built-in Julia package manager, Pkg, to find…” — better to remove the commas.
<rekado_>“and the work behind it” — this is ambiguous. On a first read it sounds like “how much work people put into the package” and not “what work the package manager performs on behalf of the user without them knowing”
<rekado_>— or maybe that’s exactly what it’s supposed to mean :)
<rekado_>the sentence on Dune is really lonely.
<rekado_>not sure what to do about it
<civodul>true
<rekado_>“NSF funded” –> “NSF-funded”, I think
<rekado_>(but I’m not sure)
<civodul>i'd put a hyphen too
<rekado_>“It extends reproducibility to the transistor level and for that reason generates interest from the Bitcoin community”
<rekado_>
<rekado_>this sentence has a distinct Pjotr flavour :)
<rekado_>to me it’s a bit jarring to see Bitcoin mentioned out of nowhere
<rekado_>and the transistor-level claim seems a bit bold
<civodul>yeah, i agree...
<civodul>i made it a bit milder than it was before
<rekado_>“on a single dye” —> “on a single die”; the other spelling is for hair
<civodul>noted
<civodul>but well, diversity is good :-)
<rekado_>the sentence that follows Bitcoin is also … visionary; I think on its own (without the preceeding Bitcoin sentence) it would feel a little less … grand.
<civodul>:-)
<rekado_>parametrize –> paramet[+e]rize
<civodul>i think the two of us just lack that sort of vision, we need to learn! :-)
<rekado_>:)
<rekado_>“Thanks to out-of-band access we can completely (re)install machines remotely.” — I’d remove this sentence.
<rekado_>it sounds like it fell out of time, because that’s incredibly common.
<rekado_>“pure GNU Guix nodes” – is that “nodes running Guix System”?
<civodul>right, noted
<civodul>yes
<rekado_>“specific software on our own guix channel” — using monospace for “guix” here is a little odd. How about just “Guix channel”?
<rekado_>“the strategy holds its promise” — I think this is actually wrong
<rekado_>“hold promise” is an idiom to say that something has potential
<rekado_>“keep its promise” would be that it is indeed as good as we had hoped
<rekado_>but maybe it’s better to get rid of the whole promise thing and rewrite that sentence a little
<rekado_>that whole sentence reads a bit too much like it was brought up in a casual conversation
<rekado_>the last two sentences could perhaps be replaced with something like: “We aim to eventually replace as many of these deployed machines as possible, adjusting Guix system services and implementing new ones as we go, benefiting the wider community.”
<rekado_>on fourth thought, perhaps the Articles section is the right place after all to mention the PiGx SARS CoV2 preprint.
<rekado_>it’s not fundamentally *about* reproducibility or Guix, but *PiGx* is.
<rekado_>“GNU Guix is a collaborative effort, receiving contributions from more than 90 people every month” — does this include the extra channels?
<rekado_>should we at least mention our extra channels like guix-science, guix-bimsb, guix-hpc, etc?
<civodul>ah yes, i guess we should
<civodul>re PiGx, i think it makes as much sense as the reproducible experiments of my colleagues
<civodul>your call!
<civodul>i think i'll stop here for today but i'll get back to the log to make these edits
<civodul>(or you can make them if you want)
<rekado_>I’ll take a little break and have dinner, but later tonight I’ll make an edit to the GWL paragraph and add the PiGx preprint.
<rekado_>I’ll also read the skipped section then and fix another typo I saw there :)
<rekado_>editing is fun. Sometimes.
<civodul>heh, thanks for your attention to detail, much appreciated!
<rekado_>I pushed a few more changes
<civodul>tx!