Community
Chat Logs
Sunday, October 18, 2026 (America/New_York)
- * phaleth is learning inkscape
- phalethflog docs are done exactly as I described yesterday, all template literals including css
- phalethI mean css in js is bad, but still
- bluehi phaleth
- bluehttps://repopack.com/project/flog/tasks/76
- blueI got it done yesterday, just forgot to merge
- phalethok, great
- bluephaleth: https://repopack.com/project/rcompat/tasks?status=open
- bluesee 57 and 58
- bluercompat is getting very close to be dependencyfree
- bluejust two more deps needed to eliminate
- phalethcool, noticed you removed crypto, crypto in JS is dumb
- blueit was just a weird package with one convenience function
- bluefs has FileRef#hash too
- blueand crypto would be shadowing globalThis.crypto
- blueso there's that
- blueanyway, it's gone
- phalethnice
- blueclanker did a full audit of rcompat, found a few things
- bluehence this
- bluehttps://repopack.com/project/primate/tasks/226
- bluephaleth: fyi: repopack *still* consumes 200 <= x <= 300 MiB of memory. Dog will specifically require to lower that number to <= 100 MiB. I don't know why it needs so much memory to be honest, is all of it JIT?
- phalethit's cause it's a VM
- blueyes I know it's a VM, but repopack isn't doing much
- phalethgolang runtime takes less memory only cause it's native
- bluethe repopack build/server.js is 36M
- phalethyeah, but you know there is event loop and that just in a sense heats up the VM
- bluewhich is already a lot and probably needs optimisaing. but 36M -> 300M is only 10x
- phalethall VMs are like that, node is actually doing very well for a VM
- bluealso bun was leaking like crazy
- phaleth36M means it should take ten times the memory, yeah
- phalethbun is a different story
- bluebun is dumpster fire
- bluephaleth: https://repopack.com/project/primate/tasks/228
- blueI'm gonna merge this, but the website itself uses the .svg as a favicon..
- bluejust fyi
- phalethnot really I think
- blueI added a <link tag to app.html
- phalethnope
- phalethno reason to do that anyway
- phalethunless you need multiple favicons on the site
- blueit didn't load the favicon.ico otherwise, for some reason
- blueit does it on primate.run, didn't do it for rcompat
- blueI couldn't be bothered to figure out, and implicit things are weird
- phalethyeah, it's fine as is
- phalethimplicit is fine
- bluewell if you can figure out why the browser didn't issue the request, then remove it again
- bluedreamreal said he saw no favicons on rcompat.repopack.app
- phalethit will load, don't worry :)
- phalethseems like flog is node specific https://upaste.de/raw/Trr
- bluelet me see
- phalethapps/website/static/favicon.ico | Bin 15406 -> 5430 bytes
- bluewow phaleth
- bluebtw, clanker said that this isn't about flog. the postinstall specifies "node install.js"
- blueflog just ran the script
- phalethyeah, I saw node in those package.json scripts I think
- phalethbut that's weird
- phalethwhy can't flog do whatever install.js is doing?
- bluewhat is this install.js?
- bluelike, where does it come from right now?
- blueI can't see it in primate website
- bluepackage.json
- phalethyou can search for `node --conditions=@primate/source`
- phaleththere are alot of those
- blueyes
- phalethactually three of those in apps/website
- blue "start": "node --conditions=@primate/source ../../packages/primate/src/bin.ts",
- blue "build": "node --conditions=@primate/source ../../packages/primate/src/bin.ts build",
- blue "serve": "node --conditions=@primate/source ../../packages/primate/src/bin.ts serve",
- blueyou mean these?
- blueyes, this is currently only necessary because package.json's "bin" field doesn't support conditions
- phalethok, that is where node is called
- phalethso node is hard dep
- bluewe're doing straight from source run, not compilation to JS
- blueno compilation*
- phalethok, well, the error message I posted said it happens on flog install
- phalethso those node calls are possibly not to blame
- blueyes, but where does that `node install.js` come from?
- phalethesbuild postinstall
- phalethhttps://github.com/evanw/esbuild/issues/4288
- nevetpost-install documentation · Issue #4288 · evanw/esbuild
- phalethtried to add --ignore-scripts to flog install, same error
- phalethtake a look at the pnpm-workspace.yml it had a directive to ignore esbuild scripts
- phalethand I think there was also some other package
- bluelooking it up
- blueyeah, oracledb was same
- bluephaleth: flog install --workspace apps/website --prod --ignore-scripts
- phalethtried exactly that
- blueare you running current flog?
- bluealso, can you paste the current containerfile that's failing?
- blueso I can have clanker test
- phalethis there newer flog released today?
- bluenot today
- bluewell there is, but it's just triggered by the website updates
- bluejust take the latest, build 65
- phalethok, I will just put that new version in and write task
- bluehttps://repopack.com/project/flog/jobs/1/runs/65/artifacts/flog-0.0.0-dev.07a7aad7035f-linux-amd64.tar.gz
- bluethe best is if you have a reproducible test case for me as a containerfile
- bluethen I can run it past the agent
- phalethyup
- bluephaleth: https://repopack.com/project/rcompat/tasks/57
- bluelook at this beauty, we're beating busboy :)
- phalethwith latest flog --ignore-scripts seems to work, but there is another error
- bluewhat error?
- phalethon build about node --conditions=@primate/source
- phalethI'll create a task
- dreamreal*sigh* two new pieces underway for BCN
- phalethI think it's a primate problem, not flogs
- dreamrealI need to get critical mass enough that other people write for it, because I'm burning myself alive
- dreamreal... although... okay, question time: blue, you read BCN because I force you to, phaleth, do you?
- bluephaleth: flog is pristine, so probably primate
- phalethdreamreal: sometimes, when I have time to lurk
- bluedreamreal: well, I read BCN for particularly two reasons. the first, I don't need to look around prefiltering what any idiot writes on his blog, I get the rough gold nuggets (usually) and 2) I get them in a REFINED form, so the rough gold nuggets are made pristine
- dreamrealWell, here's the question: am I setting the bar too high for people to be willing to write on BCN?
- dreamrealLike, I'm trying to write well but at full resolution
- dreamrealand if people go "I can't do dat, I rite gud on twitter insted" ...
- bluepartially, yes
- bluethe bar is pretty high because the takes are often complex and require thinking
- blueI do wish you had more plebs-oriented takes; like the current value destroying that andrew-zig has been hard at work on, for the last few months
- blueyou did have a take on zig's AI policy; but it was *way* too nice
- bluezig's AI policy, combined with lashing like crazy at bun, combined with the crazy and irrational move to codeberg... just shows you what a deranged dictator-for-life can cause for damage, in a project with otherwise high value
- dreamrealWell, BCN's editorial policy is to keep criticism unpersonal, and there's no way I'm digging in against anyone.
- blueI'm actually less annoyed with the AI policy, less annoyed with the childish lashing out at bun, than I am annoyed at how condescending and insufferable this guy was against *donors* on github
- dreamrealI'd like to have more plebe-oriented stuff as you call it, but I don't really think that way so it feels scripted
- dreamrealand since I don't use zig I don't have enough visibility into the specific issue, if you will
- bluewell, *you* draw the consequences, one of your most seen pieces was the bun rust rewrite, and it was closest to plebs content
- dreamrealI wouldn't mind publishing something written well
- blueI'm not gonna try and tell you how to run BCN. But if you want more exposure, and consequently perhaps more authors, you might have to stoop a bit
- blueyou don't *have* to make a zig 'hit' piece a personal attack
- dreamrealActually, I need to check the actual numbers: the bun rust rewrite piece maintains the most temperature but that doesn't actually correspond to "most hits"
- blueit is a personal problem caused by the project leader, but it is equally a project problem, by now
- dreamrealI don't mind, but I cannot stand ragebait, so I'm very sensitive to hosting it
- bluecodeberg regularly doesn't work properly; a ton of work was left unfinished on github in terms of issues, that the zig project did not bother migrating
- dreamrealRight now I'm binding pele into BCN, which will be a content piece as well, AND I'm working on a piece on irrelephants
- phalethcelld is the current hype, and the deno team being bought by cloudflare to work on celld
- blueand people are wary of using zig on high profile project, because if they pivot away, they'll get lashed out at
- dreamrealblue: See, I don't have those perspectives to even work with
- dreamrealwhich kinda sucks: I can't know everything and (surprise!) I don't
- phalethI mean clanker on celld, inlcuding the original author of node Ryan Dahl (think I spelled that right)
- dreamrealpele installed on server! Now working on one of the front ends to test it out :D
- bluebecause personally the lack of major 1 in zig is currently the biggest problem
- dreamrealblue: you mean major 0?
- blueno, I mean major 1
- dreamrealI mean, I agree with you, and you know why
- bluezig was not meant to be a 0ver
- blueand for a change, I don't think it should be. I'm in the 0ver-camp, but not for programming languages
- dreamrealSee, I know who should write this, and who should edit it
- dreamrealor I THINK I do
- blueI can tell you that such an article will generate A LOT of interaction
- dreamrealI'm sure it would
- bluebecause I know how people feel, and they feel strongly, either way
- * dreamreal actually does not, which is part of why he shouldn't write it
- blueit touches on all the current plebs seams. AI, semver, hosting platforms
- dreamrealTo me, zig is very much a storm in a teacup
- dreamrealI know it's actually NOT - that's the point of the irrelephant article I'm working on - but to me it is
- bluedreamreal: I'm just gonna tell you that if *I* wrote it, it'd probably be titled "The Zig project is going nowhere"
- dreamrealhell, write it that way! We'll edit it into shape if it needs it.
- dreamrealHideki Idoru has only published one thing so far!
- * dreamreal giggles
- bluetrue
- blueunfortunately I'm currently busy with Dog
- dreamrealI'd actually prefer a different pseudonym, or under your name, but eh, whatever. If we want to reuse it as an anonymous id, we can.
- dreamrealSit, dog!
- dreamrealLie down!
- dreamrealRoll over!
- dreamrealPlay dead!
- blueIt's edging on V8, believe it or not
- bluewhether it can beat it or not, remains open
- blueI'm skeptical, but Dog seems to be confident
- blueI'd be actually personally ok with getting lim(v8) rather than surpassing
- dreamrealdang it, now I want the zig thing too
- bluehaha
- bluethis is where we are currently at: "recent Dog timings varied roughly 59–76 µs/batch versus V8’s 3.3–4.0 µs/batch."
- blue17-18x roughly, on a non-representative benchmark
- bluewe started at 30x
- blueWe're still doing work for reference-bearing JIT, but it's not enabled yet
- * dreamreal is so jazzed by pele - BCN is getting mermaid graphs in posts, I think
- dreamreal(I say "I think" but it's actually looking like it's working on the server, but the UIs need to respect the directives)
- blueya, I recently added mermaid support to repopack
- dreamrealpele.run!
- dreamrealpele > mermaid so far, since it actually generates SVG differently
- blueusing the "mermaid" npm package, I believe. you should render it on the client, cheaper
- dreamreals/should/could/ but mermaid's expensive
- dreamrealthe problem is that we'd have to pass the mermaid block to the client
- dreamrealI AM rendering drafts on the client, though
- bluepele is great. let me replace the mermaid dep with pele in repopack
- bluePele 1.6.0 replaces Mermaid 12.1.0
- blueand its diagram dependency graph, including KaTeX, Chevrotain and lodash-es.
- blueso nice, dreamreal!
- dreamrealI'm impressed as all hell
- dreamrealGot a crackerjack post underway on it but want to have it deployed, so the post on pele uses pele :D
- blueand the zig post?
- bluehas it materialised yet?
- dreamrealHmm,I don't think I'm the one to materialize that one, it could be written here with !article (that's why that command exists, to allow people to discuss content that creates an article) but I don't think I should sole-source it
- dreamrealbut to answer you directly: no, I have two high-effort pieces underway, a third would be a lot for a calm peaceful sunday
- blueI can give clanker a few pointers, see what it comes up with
- bluebut I won't be able to read it aloud!
- dreamrealThat's fine, I'll narrate it
- dreamrealthe audio isn't literally narrated, the audio is a separate piece of content and should be considered as such
- bluefair
- dreamrealWhat I do for the audio is have the *rendered draft* in front of me, and it's mostly a mostly-faithful way of how *I read it*
- dreamrealso no footnotes unless they occur to me as I read the actual main content, etc
- dreamrealthe main thing about using !article is that if you use the AI features for it you get haiku as the model
- bluezig article is cooking, dreamreal
- dreamrealooo that's awesome
- phalethdiagrams will finally properly display in librewolf? https://repopack.com/project/repopack/tasks/701
- dreamrealI am SO jazzed about this pele thing
- dreamrealphaleth: so what did you see in the ui-primate OCI image?
- phalethnot yet, I will log into github, one sec
- dreamrealno hurry
- bluephaleth: maybe
- dreamrealI'm doing a general site refresh this morning thanks to pele but most OCI updates like that barely affect traffic
- phalethdreamreal: don't see any notification
- bluephaleth: https://repopack.com/project/rcompat/tasks/58
- bluebeating ws as well now
- dreamrealphaleth: you're on there as "phaleth" right?
- dreamrealsays it's got a pending invite :/
- phalethlooking at the Dockerfile and there are few changes I am gonna suggest in the PR, the one big mistake I see is to use npx to run the app, that results two node processes instead of one I think
- dreamreal(graalvm's native images do not do image libraries well)
- dreamrealphaleth: *nod* I used AI to build the app, I do not and did not expect it to be good at it, just sufficient
- phalethbut I will create a PR
- dreamreal*nod* and thank you VERY much
- phalethno problem
- bluedreamreal: discord, zig article
- dreamrealgot it, looking, will get to it ASAP but drowning right now
- blueget a LIFE VEST
- phalethanother big mistake is to use main as branch name :)
- blueYES
- bluefinally someone calling it out
- phaleththe sole reason why dog exists
- blueyes
- bluefinally someone who gets it
- dreamrealphaleth: I don't care, it's github default and I saw no reason to give a single crap about it :D
- dreamrealI don't care about main/master, use master and I'm fine, use main and I'm fine, no reason to get my knickers in a twist about it
- dreamrealAdding fonts and styles for primate, and getting the preview renderer in place
- dreamrealand doing a rare nextjs update so the fonts render in case any damn fool uses the nextjs front end :D
- dreamrealblue: https://primate.bytecode.news/posts/2026/10/test-post
- blueHOLY HECK THE ZIG POST IS GOING PLACES AND THE EDITOR IS BUSY FIDDLING WITH GRAPHS
- dreamrealblue: the editor has a queue and is human and stuff AND the author hasn't granted access to the zig post anyway
- blueI sent you a glink
- dreamrealyes, and I requested access because I didn't get granted access despite the link
- dreamrealokay, one of four done in the queue, now to work on #2
- blueDONE
- dreamrealthere's #4!
- bluelol
- dreamrealokay, how do you want to do this
- dreamrealdo you want to post it yourself, or do you want the same process as the prior one
- blueto what are those questions tending
- bluedo you want to draft it on BCN?
- blueotherwise I'd say, let's finish it first on gdocs, then decide
- dreamrealWell, I mean is it a hideki irodu-type "a user sent this in" post, or what
- dreamrealokay, I 'll file comments, some will be to me, some to you, okay?
- blueidk, I haven't decided yet
- dreamrealThat's fine, it just affects voice
- bluesure
- dreamrealif it's you or attributed to a pseudonym, some of the stronger statements get to stand
- blueI'm currently trying to make it sond more human, because it screams clanker, and it annoys me because I don't like reading clanker content either, feels so unreal
- dreamrealThe way to do that is for you to read each sentence and phrase it yourself, including flow, FTR
- dreamrealbut I get it
- blueyes
- dreamrealOkay, I'm goingto wait on the bun section until you're done editing
- bluek
- bluethx
- dreamrealLMK when, going to work on the pele post now
- bluek
- bluedreamreal: done
- dreamrealOkay, almost done with the pele edit, will switch when I get closer on this, won't be long
- phalethdreamreal: I found the current deployment is very insecure, there are all node_modules in the container at runtime with write permissions
- dreamrealphaleth: *nod* feel *very* free to fix
- dreamrealI make no claims to the primate ui's quality
- bluephaleth: did you see the native gains on `ws`?
- bluethis is amazing. Getting rid of a dep, and becoming faster
- phalethdreamreal: PR is already there
- phalethblue: I saw some 2x improvements in some table
- phalethwell, consider using websockets instead of SSE for repopack
- bluethis is amaze sauce
- dreamrealthanks for notifying me, GITHUB YOU PIECE OF SHIT
- dreamrealsorry
- * blue points at repopack
- blueactually, there's no way to subscribe to a project's tasks... yet.
- bluephaleth: we need a task for this
- dreamrealphaleth: deploying now
- dreamrealmuch appreciated
- dreamrealphaleth: ++
- nevetphaleth now has karma of 1.
- phalethblue: btw, should be tricky to make flog work against bcn's primate frontend, I tried nub but nub failed
- phalethyeah, I wonder what happened to the release candidates task in repopack, forgot the number
- dreamrealdefine sesquipedalian
- nevetdefine sesquipedalian: sesquipedalian (noun): A long word. (Wiktionary via FreeDictionaryAPI.com, CC BY-SA 4.0: https://en.wiktionary.org/wiki/sesquipedalian)
- dreamrealokay, queue item #2 done! back to #4 for a bit
- dreamrealblue: okay, my thought: this needs to be attributed to an outside contributor, either through authorship (you submit it) or direct attribution (as we did with the idoru article)
- dreamrealIf I narrate this, I'm gonna have to use a ring modulator or something to change the narration voice, becuase these aren't my words.
- bluetrue dreamreal
- blueyou don't *have* to narrate it though
- bluewe should make a podcast recording out of this article really
- phalethdiagrams now display correctly in librewolf
- liolocblue: Hey blue, are websockets up to date in the primate documentation?
- lioloci only found this: "https://primate.run/guides/responses/websocket-chat"
- phalethtake a look at https://rcompat.repopack.app
- nevetrcompat — JavaScript standard library
- phaleththere is currently only this change pending https://i.imageupload.app/43b5fc35da9bbdc36d38.png
- liolocoh interesting, do you know if blue is going to update the documentation website for primate.run? It looks homeless compared to this one, looks clean
- phalethyou mean the design or the content?
- liolocdesign
- phaleththere's gonna be some pema section when 0.43 is out
- phalethoh, hmm, dunno
- phalethcan you create a task in primate repo?
- liolocI can, but i'll have to first identify the issues with it, there's a few things that break on mobile
- liolochow are you btw? thanks for helping me out on a sunday :D
- phalethok, I think blue fixed some mobile issues, but it's all stuck with 0.43
- phalethI'm doing alright
- phalethcan you test against https://primate-master.repopack.app ?
- phalethhmm, I think I got the wrong url, one sec
- phaleththe server has died
- phalethbye bye blue
- lioloc😵
- phalethheh, weird, that never happens, but the uptime is of course not 100%
- liolocits ok, i'm busy doing other stuff anyway
- lioloci'm building a meal prepping app with a primate server as my backend and native swift app, personal app for now
- liolocwould be so awesome if i could have used primate for the ios app as well
- phalethcool
- phalethlioloc: https://primate-master.repopack.app is the one that tracks master branch, if you find issues there then please create tasks
- nevetPrimate — The Universal Web Framework
- phaletheven small tasks are good, does not have to be super detailed
- bluephaleth: hey, did you restart the server?
- phalethblue: there was a server restart all the sudden
- blueah ok, hm contabo
- phalethactually a slow one, but contabo never claims 100% uptime
- phaletheverything needs maint
- phalethone container didn't start, I think that's repopack docs, will redeploy it
- blueright
- bluelioloc: you had a question?
- phalethhmm, it only took 7 seconds to deploy
- blueyes
- bluephaleth: I merged the rcompat native ws rewrite, so rcompat should now be completely depsfree and ready for dog
- phalethok, did you merge the favicon and logo optimizations?
- bluenot sure
- bluenot yet, will review
- blueI like the rcompat logo
- bluedunno who made it, but it's cool
- phalethyeah, but the SVG is huge, so I made it simpler
- blueI think it's time to start looking at repopack#283
- blueI've been pretending it doesn't exist for a while
- phalethtrue
- phalethhttps://repopack.com/project/rcompat/tasks/59
- blueyes, reviewing
- bluethe rcompat image builds in 15s.. still too slow
- phalethI'm working on logo and favicon for flog
- blueand dog?
- blueremember dog's motto...
- phalethrcompat build uses pnpm
- blueI guess flog comes first because it already has a website
- blueyes I know, it needs to switch to flog
- bluedog's website will be dog.flog.run
- phalethI was planning to do that, but still kinda waiting for trust me bro installer
- blueyes that's 283
- phalethnah, dog does not need a website yet
- phalethalso flog's docs are outdated already
- blueoutdated how
- phalethwaste of tokens, it should have been just a frontpage
- bluethis reminds me flog needs `flog outdated`
- phaleththere is something about primate migration
- * blue creates task
- phalethheh
- blueyeah, the primate migration needs to get out of there
- blueclanker overcomitted on that
- blueI somehow don't like rcompat's font
- bluesuch a generic documentation font :/
- bluefeels like it's inter of smth
- blueor smth*
- blueoh, it is inter, LOL
- bluetotally soulless
- blueI like the code snippets
- blueboth the colours and font, ui-monospace. they're good
- blueI just don't like the prose font
- bluemaybe you have a better suggestion
- phalethlioloc seems to like rcompat docs
- phalethI only need to see if the logo gets updated properly
- bluewell he helped writing them, why wouldn't he like them
- blueor do you mean the website
- phaleththe design
- bluercompat logo merged
- bluehttps://repopack.com/project/primate/tasks/226
- blueI'm honestly thinking of pulling this into 0.43
- blueI have half a mind to release a primate 0.43 that comes with totally outdated rcompat versions, and then I'm gonna need to start backporting nonsense
- blueso better pull off the plaster now than later
- phalethcool, looks fine
- blue(that's bandaid, for our american friends here)
- liolocblue: hey blue! yeah i was just wondering if the docs had a page on websockets, or if the most up to date page was the guide "https://primate.run/guides/responses/websocket-chat"
- phalethyeah, rcompat updates is just chores
- bluehey lioloc. so are you asking if it properly reflects the source code, or whether it updates with 0.43?
- liolocand yeah i prefer the documentation website for rcompat than primate
- lioloclooks nice
- bluebecause we haven't done a lot of work on websockets in 0.43
- liolocmore so if it is up to date in general
- bluewell, the @primate/http example is a bit ridiculous, we should switch to marko or svelte, but I thik it's uptodate
- liolocin case it wasn't an old page
- bluelet me sweep over it shortly
- blue@primate/html*
- bluethe guides should ideally contain switchable examples when they show frontend snippets, but it's too much work
- bluemaybe now with AI it's improvable
- bluebtw lioloc, I'm not particularly set-in-stone about the primate website design, if you have another suggestion, create a task and send it to revie
- bluefortunately now with task apps I can quickly review for aesthetics
- blueand yes, the example is correct. it contains a small bug, which I'm going to fix for 0.43
- liolocyeah tasks make it quite easy to make updates, pretty neat
- liolocoh it does? I'm using it atm
- liolocnot with html
- liolocjust as a raw endpoint
- blueit's a huge bug, just a small incorrectness in the examples, I'll show you the fix
- bluehttps://repopack.com/project/primate/tasks/230
- blueit's not a huge bug*
- bluelioloc: the next rcompat versions got rid of their last few deps (busboy and ws, both for @rcompat/http on node), and replaced them with custom implementations, which are faster than those packages
- bluercompat is now completely dependency free, as it should be
- bluethe stdlib should not rely on npm
- bluethis is also somewhat relevant for the future of rcompat and flog/dog https://repopack.com/project/flog/tasks/74
- blueI believe I found a golden path there, but it's yet to be seen if it's doable
- bluethere's also `flog format`, `flog lint`, `flog test` (which is proby, really) and `flog test --coverage` coming for flog, but most of those need AST stuff and thus will need to wait for dog
- blueflog will refuse to publish with any "std:" imports in the tree, unless --std-compat or --std-pure are provided
- liolocwhen you say rcompat is completely dependency free, it still uses node for some stuff or not at all?
- blueyes, it uses node, just not npm
- liolockk gotcha
- liolocthat's insane
- liolocvery cool
- blueya
- liolocand you're using flog already?
- blueyes, primate has already switched to flog for 0.43, and it's faster than pnpm
- blueand safer, too
- bluewe're waiting for release primitives to land in repopack so people can download flog from flog.run
- liolocneat
- bluewell, 'waiting' is a general term here. I need to do it
- blueit's more like phaleth is waiting and I'm wasting my time playing with Dog
- bluebut I sometimes need the right mindset to get into a task, and until I get it sorted it in my head, I don't start
- blueright now it's a bit hard to conceptualise whether releases draw from job runs or vice versa
- blueI guess they need to draw from job runs, but also that means you can only create a release after the job has finished running, at the ref you want to create the release
- bluemaybe we'll just go the gh way where you create or choose a tag. but still creating a tag would then need to run a job and create the artifacts first
- blueanyway, I'm not entirely convinced here which way is the good way, need to sit on it a bit longer
- blueshould be done hopefully by next week, I want to push 0.43 out the door
- phalethcan have release stable and release unstable
- phalethstable is supposed to be tagged
- blueyeah but like, I don't wanna a second mechanism to create artifacts
- bluejobs already do that
- phalethone stream of runs, overlaping
- blueand releases should probably include the source in one way or another
- phalethsecond mechanism?
- blueyes
- bluejobs can already create artifacts
- bluebut releases should offer the source as tar.gz etc.
- bluethat's another 'artifact'
- blueI don't want two systems
- bluemaybe it's unavoidable
- phalethcreating tag does not have to run job, just take artifacts from tagged commit and rename those
- blueyes but who creates the tar.gz of the source at the ref?
- bluewhat primitive? the release or the job run?
- blueif the release owns it, then that's two mechanisms that own artifacts
- phalethI'd not worry about tar of source
- phaleththat's like duplicating the source code, downlodable artifacts is the goal
- bluefair, so you're leaving it to the jobs
- blueyou know, maybe jobs should just have the option of delivering source artifacts
- blueI think that possibly solves the conundrum
- blueand then releases are just light tags that link to job runs (and freeze them, so they don't get cleaned up)
- phalethtar.gz is another issue, that needs to be downlodable on demand
- phalethnot stored permanently with release
- bluek
- bluephaleth: there's a new feature on repopack
- blueif you hit p, you get the quickjump to projects, don't know if you knew that
- bluehowever, if you fully type the name of a project, you can also quick-quick to a task there
- bluetry it, hit p
- bluethen write repopack:283
- blueand Enter
- phalethI knew about /
- blueyes but you didn't know about p, t, j and a
- bluehitting t, j or a and immediately Enter also jumps to the section itself
- bluekinda neat
- phalethforwardslash does not accept repopack:283 but p does, weird
- blueyes because forwardslash isn't in project mode
- phalethah, you can switch to /projects from /
- blueyes
- phalethblue: take a look at tasks in review in the flog repo
- bluedreamreal: I did another sweep of the article. I think it's much better now, care to rereview?
