Community
Chat Logs
Sunday, October 25, 2026 (America/New_York)
- liolochey blue!
- phalethant migrated to upm https://github.com/theMackabu/ant/pull/143
- nevetmigrate from libpkg (zig) to upm (ts) by theMackabu · Pull Request #143 · theMackabu/ant
- phalethnotice how many lines of code got dropped
- bluehi lioloc
- bluephaleth: I know neither libpkg nor upm... but ant is a VERY badly chosen name. there's a buildtool for java called that way
- phalethflog 0.1.3 can build bytecode.news site
- phalethblue: naming is hard
- bluewow, flog is amazing
- phalethyeah, the build of the image is slower with flog than with npm/npx by about 34 seconds cause of some dev deps that have to be installed
- phalethupm looks interesting
- phalethand it's also nice that ant was able to remove all the zig code and get better compatibility with windows
- blueflog 0.1.4 is out
- bluephaleth: is flog is slower than npm, then I won't replace the BCN stuff. flog must be faster than other package managers in all scenarios
- blueif flog*
- phalethwhen `flog lock` is executed does it convert `package-lock.json`?
- phaleththere seems to be no equivalent to `flog ci`
- blueno, flog lock currently doesn't convert any existing lock files
- phalethok, the bcn deployment uses `npm ci`
- bluethe flog equivalent is `flog install`
- blueflog install never modifies a lockfile, it just materialises it
- phalethalso `flog install --prod` fails to install some shacdn whatever which makes `flog build` fail next
- blueok, let me see if I can see why
- phalethyeah, but the flog.lock is missing and there is only package-lock.json
- blueyeah, that's why you run `flog.lock`
- blue`flog lock`*
- phalethok, let me send you the Containerfile
- phalethbut flog.lock does not convert so that means it will just look into npmjs.com for versions
- blueonly when you do `flog lock` the first time. otherwise you commit the flog.lock file alongside
- blueui-primate on master [$?] via v26.11.1 took 5s flog install
- blueinstalled 616 packages and 0 workspace links (143 cache hits, 473 downloads)
- blueui-primate on master [$?] via v26.11.1 took 9s flog install --prod
- blueinstalled 138 packages and 0 workspace links (138 cache hits, 0 downloads)
- blueselected 0 source workspaces
- bluelooks good no?
- phalethblue: I've only added flog install steps to this Containerfile https://upaste.de/raw/r3z
- phalethah, so you have flog locally
- blueya but that Containerfile is unfinished
- blueit needs to copy flog.lock
- phalethoh, there is flog.lock in the repo?
- blueand drop package-lock.json
- bluethere isn't, but there should be, as part of the transition
- phalethso flog is not going to support package-lock.json?
- phalethok
- bluenot that I intend to
- blueI mean, you either use flog or npm to manage the repo, I think
- bluemixing is bad
- phalethyeah, but how to adopt flog here
- blue`rm package-lock.json && flog lock`
- bluethat's a one-off migration
- blueafterwards the Containerfile can be adapted to copy flog.lock
- phalethok, let me see if that will make the build faster
- phaleththere is also `.npmrc` which I have no clue what it does
- blueit contains this
- blue@streampack-dev:registry=https://nexus.streampack.dev/repository/npm-group/
- nevetRepository - Sonatype Nexus Repository
- blueI have no idea what that is
- bluepossibly for those two deps:
- blue "@streampack-dev/clientjs": "^0.2.3",
- blue "@streampack-dev/masthead": "^0.5.5",
- blueidk, we're gonna have to ask dreamreal what this sonatype nexus shenanigans are
- phalethlooks like custom npm repository to me
- phalethdoes flog support that?
- bluedon't think so
- phalethI think otherwise one would have to specify something like: "https://nexus.streampack.dev/repository/npm-group/clientjs": "^0.2.3", in package.json
- bluewell the question is how do those custom repositories work
- blueis it only for certain packages, or all, or?
- blueI'll investigate
- phalethwell, they should have the same API as npmjs.com
- phalethyou remember the move of sheetjs outside of npmjs.com to their own custom npm repository
- phalethor registry or whatever that's called
- phalethand npm as client can handle that
- phalethI've used sheetjs in a project in the past, let me try to find it
- blueya
- phalethlooks like sheetjs changed the installation process and now explain how to vendor the dep, but there's still some doc https://docs.sheetjs.com/docs/getting-started/installation/nodejs#legacy-endpoints
- nevetNodeJS | SheetJS Community Edition
- phalethwell, I've tried to provide flog.lock to the bytecode.news project and running into same restrictivity as I've ran in past: flog: error: flog.lock does not match the project: lockfile does not represent current target linux/x64/musl (represented: linux/x64/glibc); run flog lock after reviewing manifest changes
- phalethError: building at STEP "RUN flog install --prod": while running runtime: exit status 1
- phalethcause of use of alpine container to do the install
- phalethtask is already there
- phalethusing flog.lock for bytecode.news frontend does not make the installation faster
- phalethI mean the overall container build that now has to use debian slim
- phalethcontainer image build*
- phalethbut anyway, big milestone for flog
- blueya
- bluephaleth: https://www.npmjs.com/package/@rcompat/fs
- blueflog published its first package
- bluedreamreal: can you make two corrections to the zig article?
- liolocblue: hey blue, the new @primate/native package is not yet released right?
- phalethblue: nice
- dreamrealblue: I can but are you not able to edit it?
- bluelioloc: hey, not yet
- bluedreamreal: no
- blueI don't see an edit button
- dreamrealokay, I think I know why - decision about people editing post-approval
- dreamrealwhat edits
- blue11:28 <phaleth> here I think the words stop and start should be switched https://i.imageupload.app/93d34153359774c87400.png
- blue11:29 <phaleth> and here it seems a sentence is unfinished https://i.imageupload.app/a6021d7054a5ff94be58.png
- blueso for the second one, phaleth suggests "anything but appreciative", which I agree
- bluedreamreal: ^ + ^^ + ^^^
- * dreamreal sighs
- dreamrealOkay. I do not approve, but I'll make the changes.
- bluewait wait
- bluethe first one is a legit error
- blueand why do you not approve of the second one?
- dreamrealBecause they were your words
- bluedo you prefer the elliptic style?
- dreamrealI actually disagree with the first one, because it's valid either way semantically
- phalethheh
- dreamrealand the second: you wrote it. Own it.
- phalethit's immutable like repopack comments
- * dreamreal shrugs. It's not. It's a matter of how you view your creative work.
- blueI guess you can read it either way
- bluefine, leave it
- dreamrealToo late
- bluewell, now you have to implement an edit feature so I can change back it :P
- bluechange it back*
- bluebut anyway, next time let's discuss it, if you disapprove. I'm always open to hearing another perspective
- dreamrealThere already is an edit feature: it just stops once the thing gets calcified, because the site author has a particular view towards creative work
- dreamrealhaving been created, I see it as a thing itself, errors and all
- blueya, it's like repopack comments
- dreamrealI did not DISAPPROVE
- dreamrealI disgareed
- dreamrealthere's a massive difference
- blueyou said you do not approve :P
- dreamrealWhich is not disapproval
- blueok, so let's discuss it anyway next time, even if you don't disagree but only disagree, or not approve
- * dreamreal shrugs
- blueif you don't disapprove*
- dreamrealokay
- dreamrealI find the first change (start/stop) a matter of semantics and it works either way
- dreamrealthe second change sands down your voice
- phalethtoo funny
- blueyeah, and upon rereading I got the point. the first change is from *describing* reality (stop -> start) and *prescribing* it (start -> stop)
- dreamrealand as a musician, you eventually come to a place where you play what you played, and there it is: fiddling with it endlessly after the fact eventually erodes any hint of soul it might have had
- dreamrealI'll correct misspellings, most of the time, but ... honestly, my contorted grammar in my earlier writing, that was me
- dreamrealstill is, and I still write that way sometimes
- dreamrealc'est la vie
- phaleththe article has soul, that's for sure
- bluebut both changes are not something I feel very strongly about. and that's also a personality quirk of mine. there are things I prefer to not have a strong opinion about, because... I don't. so I leave room for others' strong(er) opinion, and I think that's valid. I heard phaleth's side, I think you should allow your side to be heard as well
- * dreamreal is not the author
- bluethat doesn't mean you can't have an opinion
- dreamrealI have a lot of opinions, most of which remain isolated as they should be
- blueyou're also not the author of primate, and you can also have an opinion, and make suggestions...
- * phaleth now thinks his repopack comments have a soul
- bluehaha
- dreamrealYou thought they did not?
- dreamrealask martin buber
- nevetMartin Buber (1878–1965) was an Austrian-Israeli philosopher best known for *I and Thou* (1923), which contrasts treating things as objects ("I–It") with genuine, mutual encounter ("I–Thou"). By his view, your repopack comments could have "soul" if you relate to them as a Thou, not merely an It.
- dreamrealThat's... not entirely correct but close enough
- dreamrealbut buber did have a lot to say about how people interact with their world
- bluethe problems are not dissimilar. someone who has read an older version of the Zig article might post a reply that is now invalidated
- bluethere are reasons to keep things as they are, and not purely for an "authentic snapshop of time" reason
- dreamrealThat'd be a particularly anal reader; "no, no, you start at THIS and stop at THAT, it's not the other way around where you STOP at THIS and start at THAT unless you decide you want to look at the two inversely, you know. ACTUALLY..."
- bluerepopack comments are uneditable the exact reason that discussions can always be understood linearly
- bluefor the exact reason*
- bluethat is worth the price, I think
- blueretconning things always creates danger... there's a reason they say you should never edit migrations, even if they contain nonsense
- phalethmigrations containing nonsense is guaranteed, so is development project issues
- blueya
- dreamrealI'm working on a piece about the information economy, because what I think is happening a lot is that people describe it as an economy while not treating it like one
- bluewhat is the information economy?
- dreamrealIt's not an it, for one thing, but it's a description of a model where information is value in and of itself
- dreamrealopen source shares the value innately, but people take the value and hoard it
- bluehm
- dreamrealand we end up with crap like php2shell
- dreamrealbecause people don't share back enough: "it's open source, they gave it to me, why should I give back?"
- bluethe alternative worldview, dreamreal, is that php simply sucks, should never have existed as a programming language, and we should retire it out of existence ASAP
- dreamrealand it's not just MONEY to give pack - php is drastically underfunded for its userbase, despite being utter crap - it's actual interaction and contribution, which is information and has value itself
- dreamrealNot the point, people relied on it despite its vast suckitude (I despise PHP)
- bluelike, if I tell you of some building material, which is really easy to build with, super friendly, but also it's a dumpster fire that will collapse with the slightest wind, and you build half your infrastructure, and most of your informational infrastructure with it, it's kind of your fault at that stage, and you're get your just deserts
- dreamrealMissing the point that we could learn from PHP
- dreamrealPHP has been crap since early on: it's entirely accreted with no design
- dreamrealzend tried to fix some of it but there's too much momentum
- dreamrealwould have been worse than python 2 -> 3
- dreamrealbut you see the same pattern all over the place
- dreamrealI've gotten five projects new releases this week, all from simply exchanging information with the maintainers
- dreamrealnot even "oh it's broke" but just asking questions that I couldn't find the answers to, trivially, when I thought I should be able to
- dreamrealI usually try to share what I found, along with workarounds and why
- blueya
- phalethis it like php people thought it'd be a good time to update their legacy php 5.4 projects yet again all the sudden? just give them a few days to realize they don't really know what they are thinking
- phalethI think php 8.2 and onwards is fine and also the use of composer is a good addition, even tho a bit weird
- phalethbut where did all the vendoring of deps take all those php 5 people? exactly nowhere and that's btw the majority of those 70% projects running the internet or whatever
- dreamrealphaleth: PHP's problem isn't *versioning*
- dreamrealit's design
- dreamrealit's accreted by people who didn't have a vision of the core product, so it's highly inconsistent
- phalethwell, that too, but you can say the same about JS
- dreamreal... I do
- phalethyeah, it's like every single PHP dev is using different tools, I know
- dreamrealNot just different tools, but writing a PHP app means using APis that LOOK LIKE C and LOOK LIKE other things too
- dreamrealthere's no coherent design, the core API is a chimera
- dreamrealit's just the way it is
- phalethyeah, for sure, I mean standard library is also a tool
- phalethbut anyway, people will just keep thinking this pays our bills and move on whatever the complexity
- phalethsimply shoehorn everything useful back into php 5.4 and that's it
- dreamrealor stop using php. I've still got two WP sites, and I'd love to move away from them. :/
- phalethwell at least WP is easy to update unless you are stuck with some of it's legacy plugins that you necessarily had to vendor at some point in time
- blue`flog bump` is coming
- phalethwhat's a flog bump?
- bluewell, flog currently detects release-sets, but there's no way to do it before you run `flog publish --dry-run`
- blue`flog bump --dry-run` shows you what other packages in a monorepo needs updating
- blueand `flog bump` edits the package.json files
- phalethsounds like flog check-updates
- phalethflog npm-check-updates -u
- phalethrepopack has 18551 lines of Bourne Shell in 92 files
- bluethat's `flog upgrade`
- bluebump just makes sure your release set is sane
- phalethok
- phalethrobots.txt has a cool icon on repopack website
- phalethcan I change repopack's logo and favicon?
- bluesure
- blueif you rework the logo, I can't promise I'll accept, but you can change it
- blueso far the logos have been excellent, for rcompat and flog
- phalethalright
- phalethcurrent logo is boring
- bluek
- phaleththe situation with templates in repopack project is also kinda crazy
- bluehow do you mean?
- phaleththere are two app.html files that are same
- bluelol
- liolochey blue! Do you have a target release date for 0.43?
- bluelioloc: hi, yes, today in Elon Musk time. Which means that if I'm lucky, probably sometime next week. I was *aiming* for today, but this is a big release and I need to audit it carefully
- blueI might still get lucky and manage to get it out today, but don't count on it
- liolocno worries
- blueI know you're waiting for the webview stuff, haven't forgotten about you
- liolocthanks
- bluedreamreal is also waiting for some stuff, I'm doing my best :-)
- lioloci completely understand
- blueflog 0.1.7 is coming, that should unblock some rcompat bumping, which should unlock the last primate updates before the release
- bluelioloc: btw, flog is now openly available via https://flog.run
- nevetflog — predictable JavaScript package management
- liolocoh neat!
- liolocwill try it now
- blueflog 0.1.7 is out!
- liolocFor a first project without a lock, create or edit package.json, then make the resolution step explicit:
- liolocwhat does this mean? You always have to run `flog lock` on new projects? (i deleted node_modules and i'm about to try flog instead of bun)
- blueyes, you run it the first time you start
- blueit creates flog.lock
- lioloci'll probably do an flog-init zsh alias that does that
- blueflog is a bit different from other package managers. `flog install` does not modify the lock file, it materialises it only
- blueflog add / remove do change the lock file
- blueadd is *not* an alias for install
- liolocwait so if it materialises the flog.lock file when you run flog install, why do you need to run flog lock? what does that do?
- bluethat creates a lock file for a project without one
- bluematerialising just means reading whatever is in flog.lock and installing that into node_modules
- liolocoh
- liolocmaybe a flog init would be nice
- liolocthat would run "flog lock
- liolocgit diff -- package.json flog.lock
- liolocflog install"
- liolocanyway I will just create an alias for it for now
- bluesounds good, could you create a task at https://repopack.com/project/flog
- blueflog init could also create a package.json if missing, with some sane defaults
- bluelike "type": "module"
- liolocyeah would be nice
- bluelioloc one thing that flog does which I love, and pnpm and no other package manager does, is release set. it verifies that you can't publish a workspace package in a way that would mess up its dependencies
- bluehttps://flog.run/docs/release-sets#the-failure-this-prevents
- liolocinteresting
- liolocvery cool
- lioloci think i created the task (i can't actually see anything on the dashboard)
- liolocwait nvm maybe it was just slow
- liolocyup i see it
- blueya
- bluelioloc: what is this for: git diff -- package.json flog.lock ?
- liolocits on the installation instructions, i just realised it wasn't useful either
- liolochttps://flog.run/docs
- nevetDocumentation · flog
- blueok, makes sense
- liolocI just copied pasted it blindly without thinking
- liolocbe back later
