tailwindcss is cool
#hangar-dev
devout nacelle
pallid yoke
It's not a component library tho
And tbh, i much rather have reusable classes than a bunch of what is basically inline style
random badger
hmmm. so typescript errors aren't showing up with the normal linting...
which means they aren't caught on commits. That needs to be fixed, or we will always end up having to double commit to fix any typescript errors we missed that cause the build process to fail
pallid yoke
Well, typescript errors also show up in serve output
random badger
idk why they aren't, eslint is extending the config nuxt says to if you are using typescript
yes they do, but that doesn't prevent you from committing/pushing the error
the pre-commit hook prevents you from committing linting errors but that doesnt cover the typescript ones apparently
I guess double commits to the master branch can kinda be avoided if we just do prs that require a successful build before merging, but thats kinda annoying for small fixes. It'd be good if that check was done on the developer side
pallid yoke
Well, later, yes
But for now, it doesn't matter
It also doesn't matter if we push shit to master
random badger
is there a reason not to allow spaces in page names?
regex is ^[a-zA-Z0-9-_]{3,}$ currently
we do slugify the name for url stuff so...
random badger
dang ok, I'm pretty happy with the pages stuff
and did we decide that it should be...
author/projectSlug/versions/platform/versionname?
to avoid ugly numbers in the url?
pallid yoke
I don't remember, but let's go for that for now
pallid yoke
And yes, adding a space to the page name as you did is fine too
random badger
man, looking through at the version creation code... YIKES
I'll be glad to delete a lot of that π
from back when I was just trying to copy function for function from ore
pallid yoke
Yeah so was I
And like, it wasn't a bad option back then
But am also glad we redoing everything now, lol
random badger
yeah, nuxt-frontend turned into redo-everything
pallid yoke
Oh btw, I think we don't need to make internal api accessible to the outside
If I understood this correctly, pages are always rendered on the server
So in theory most internal api shouldn't be called from the browser
I guess it's only the actions, forgot about those
random badger
hmm I dont think that's the case.. I think its the first request that is ssr
cause I use the network tab to debug requests to all routes
but can just use cors to restrict it right?
random badger
dont think we need the CSP stuff now either
cause spring has nothing to do with the rendered content
pallid yoke
I took tomorrow of work, maybe I can finally be helpful on hangar again then
random badger
Ok, version creation, here I come
pallid yoke
oh you got ton of progress, way more than it looked like even from the commits, lol
I dont have much time today, but ama try to fix one issue I found
random badger
yeah, I tend to do big commits that are full of lots of random stuff when re-writing a project rather than feature based commits
random badger
well wait, whats wrong w/that?
pallid yoke
level 1 and level 2 worked, afterthing below that added the full slug
random badger
that was my intention?
adding the full slug? makes it easier on the frontend to know what the full slug is
pallid yoke
check page test5
it is below Test2 test3 and test 4
but slug was "Test2/Test2/Test3/Test2/Test2/Test3/Test4/Test5"
because it was copying the full slug all the time
so slug of test2 + slug of test3 + slug of test 4 + slugified named of test5
when slug of test4 already contained slug of test 3 (which contains slug of test2)
I suck at explaining, lol
just try creating pages a couple of levels deep and you should see what I mean
the slug grew exponetially basically
pallid yoke
(in my screenshot, first 4 rows are the broken behavior, last 4 rows are the fixed)
random badger
yeah, all it should do is tack on its slug to its parent cause its parent should already have all the rest
yeah, the recursion was operating under the (wrong) assumption that only its slug was stored in the table
pallid yoke
that was harder to explain than to fix, lmao
random badger
yeah, that stuff is always a pain to explain
pallid yoke
at least I was helpful today
random badger
now... how do we get the Docs tab to match /pages route....
it'd be nice if it was highlighted on child pages routes
maybe just have to do that manually tho
random badger
yes, but its route is /author/slug/ not /author/slug/pages/parent/child
and we don't want it matching for /author/slug/versions
or /author/slug/settings
so the docs one kinda has to be exact
random badger
cant say Ive ever put an annotation on a type like that before.
random badger
we also need to have change how the recommended version works, so projects can have one for each platform
atm, its just one recommended version per project
should be one per platform per project
maybe a new db table?
id - created_at - version_id - project_id - platform
unique constraint on project_id and platform
random badger
ah, the big SQUASH at the end of this is gonna be satisfying
also should go through and weed out some of the issues that might not apply now
pallid yoke
Yeah, the great cleanup will be at the end, and hopefully we can publicly call for testers then
And get a bunch of new issues (bugs and requests) then π
I don't totally remember how recommended versions work right now, was that that strange json field? In that case, yes, i very much prefer a new table like you outlined
!!
Heh
random badger
Nah, itβs not json, itβs just one version id on the project table
random badger
https://github.com/PaperMC/Hangar/commit/dc9ad9c2c0f1880b365bc4a2f4a5e7d53dafff3...
Yeah... idk why thatβs there, but it is. π€·
I think the only json thatβll be left is project notes?
pallid yoke
Which we could move out too
pallid yoke
Yeah... idk why thatβs there, but it is. π€·
Well, it can be collapsed to just tab.exact, right? I can handle that after shower and stuff
pallid yoke
re todo list, we got this https://github.com/PaperMC/Hangar/issues/314
prolly can clean it up
pallid yoke
Yes!
And while at it, doing some cleanup on the backend
Some technical dep we inherited from ore
pallid yoke
It's like next
But we don't write the buisness logic in javascript
That's all spring
errant jacinth
Ohh like Nuxt handles serving the page but it uses the Spring API for data?
pallid yoke
So frontend, but mostly server rendered, which talks to spring backend for getting the data and doing actions
errant jacinth
Yeah makes sense
random badger
Yeah, itβs just way nicer to use vue to write the frontend than having to juggle vue and freemarker
pallid yoke
Added benefit of it being fast
Like, once we can get some caching into the backend, and I do a bit of seo magic, stuff will be amazing
random badger
god it feels so nice clicking around on the site w/no page loading times
insta-navigation
modest forge
hangar-new when
pallid yoke
Runaway wen?
random badger
maybe a new db table?
id - created_at - version_id - project_id - platform
so Mini, this then for recommended versions?
pallid yoke
Yeah
random badger
I'm wondering if there is a gh action that can create an issue that tracks all TODOs in the code
random badger
that second one is just checkboxes in pr descriptions right?
modest forge
yeah nvm just read it
https://github.com/dtinth/todo-actions does an issue per TODO
pallid yoke
Issue is also that all the actions only check stuff on commit
Idk, maybe I can steal some code
modest forge
I mean, if you don't commit anything, the TODOs won't go away, so does it really matter if they only update on commit?
random badger
well getting all the todos initially would be the trick there
pallid yoke
Ye
random badger
but idk if one issue per todo is a good thing... dont really want 150 new issues
modest forge
or do you
random badger
dont wanna turn into Paper π
pallid yoke
It's also not helpful
I'll write something later
That just scans the whole code
It's easy enough
random badger
π
pallid yoke
First gonna cook smth tho
random badger
yeah, ok so there are 2 places where we store json in the db, project notes, and all the comments on a project version review
not sure what the best way to handle those are
(just a note, I don't think there was even a way with old or even ore to add project notes)
random badger
ah ok
pallid yoke
ok, lets see how to do this todo scanning
random badger
I think we need to also set some standards when it comes to capitalization I think even now w/the rewrite some stuff in inconsistent regarding looking up projects by author & project slug
pallid yoke
I would say ignore capitalization for that stuff
pallid yoke
I guess, ye
random badger
probably also need to look for .equals in the code then too and change to ignore case where appropriate
pallid yoke
ye
pallid yoke
easy part is done
(just ignore this regex)
<!--[ \t]*@?(?:todo|fixme):?[ \t]*([^\n]+)[ \t]*-->|(?:@|\/\/[ \t]*)?(?:todo|fixme):?[ \t]*([^\n]+)
now to look at what to with the result
am thinking, you configure an issue, it posts all the tasks into the OP, and a comment, where stuff that got removed is archived, which we could mark as offtopic
oh also I want this to create a comment on PRs, this PR will add X todos, remove Y todos
we prolly wanna rename todos a bit π
ivory ingot
RFC: "Not Todo List"
pallid yoke
"Exception in thread "main" java.lang.NullPointerException: Cannot invoke "org.kohsuke.github.GHRepository.getOwnerName()" because the return value of "org.kohsuke.github.GHIssue.getRepository()" is null"
shit api, smh
this is fun
runic rapids
π¬ going to hangar
pallid yoke
is there a way I could hide info in the markdown? π€
a comment works, nice
github is really not happy with me, lmao
when I press edit in the interface, I get some old content
pallid yoke
this is the stuff that would be written on a PR
(removed is so large because I made the script ignore some folders and didnt let it update the comment cause I dont wanna get IP banned :D)
pallid yoke
I wish the issue itself would display the number of open tasks
you can only see it when its referenced
oh and in the overview
and I guess it doesnt matter since I write summary anyways
random badger
ok, had a thought about maybe the platform can be the last part of the url path?
Machine_Maker/Maintenance/versions/3.0 then wouldn't 404, but show different platforms with that version string?
(or if it only had one, could show that version)
then you could do Machine_Maker/Maintenance/versions/3.0/paper or Machine_Maker/Maintenance/versions/3.0/waterfall
or even could do another v-tabs situation with the 3 platforms (or however many that version supported)
this would work for uber jars that support all 3 at once OR different jars that have the same version string
oh that reminds me... need to change where it stores the plugin jar. rn its in a folder named the version string, which isnt unique anymore
pallid yoke
Or we could do it a bit different again
Where in upload, you upload x jars for all platforms
And then it's basically one release (one version, one text) but multiple jar, but one page to download them all
@barren shale, since you have plugins on multiple platforms, do you have a need for unique change logs per platform or would being able to upload all jars at once be nicer?
barren shale
at once π
ivory ingot
If the changelog is markdown we can structure it ourselves to declare specialities.
pallid yoke
What do you think about this idea jake? I think it's nicer UX wise both for authors and users
random badger
So each version could be tied to more than one file?
Do we then go back and undo everything so version strings are still unique in the project_versions table? Or should you still be able to upload a version with the same string at a different time?
I have to think more about how that would work backend wise. And itβs 1am. So that doesnβt help
pallid yoke
I didn't think about the impl, just tried to approach it from a UX Pov
And yes, get some sleep
random badger
I think there are authors who would want to do separate stuff tho. Doesnβt leaving it as it is leave more control to them?
pallid yoke
Mmh
Maybe we can leave it distinct on the backed
And do a unified release process in frontend optionally
(which we would work on later)
But yeah, keeping it separate on backend allows for more flexibility, we can always solve this on the UI later
random badger
Yeah, I say on the backend, we keep it as one version db row has one file, but maybe later we allow the frontend to create 3 versions at once
pallid yoke
Yeah
random badger
I think regardless tho, we still should decide on how the version page layout should work
/_author/_slug/versions/_platform/_name OR /_author/_slug/versions/_name which has links to each platform, or just displays the one if there is only one
random badger
sure, and thats /_author/_slug/versions/_name/?_platform
cause we can have the tabs linked to a url
pallid yoke
re: "why doesn't this work as a replace? It'd be good that this page wasn't saved in history" on linkout, history api requires same origin
random badger
So not possible then?
pallid yoke
no
random badger
now I'm second guessing what we are doing w/versions
should each version row in the db table correspond to a platform the plugin is compatible with?
its just if we are displaying each platform separately (even if its a uber jar) wouldnt that be easier?
pallid yoke
I am not sure am following
Oh and if my auth change broke stuff for hangar auth, just revert, having to reload when I use fake user is acceptable
Tbh I don't know how we store versions rn
random badger
Maaaaaaaybe should switch that to the nuxt-frontend branch?
pallid yoke
Or just disable for now, thought about that too, will look at it when am at my laptop
I played a bit around with a graph lib, since the admin pages have some graphs
Looking nice, will not be able to finish today tho, but tomorrow
sly mason
somber flicker
So dare I ask but are we any closer to Hangar going live?
I ask primarily because ....Spigot have just flaggeed anyone who uses Paperlib.suggestPaper() as advertising and removing the plugin from Spigot....
or you can update and remove it
misty mountain
read: spigot started enforcing their years old rules
somber flicker
...given its not actually advertising ...prism actually does work better with paper...
misty mountain
it is advertising under their rules, and has been for many years
hangar will be ready when its ready
somber flicker
maybe I should just change the wording......"You can use Spigot but numerous events wont be tracked. If you wish to track all events please use Paper."
soft hornet
I ask primarily because ....Spigot have just flaggeed anyone who uses Paperlib.s...
File removed or entire plugin listing banned?
somber flicker
@soft hornet they havent specified the punishment at this stage just said its a rule breach
ember ivy
imagine using spigot resources
pallid yoke
We decided to rewrite the frontend of hangar and cleanup some backend stuff, but we are making fast progress
misty mountain
please direct any and all rule discussion of spigotmc to phoenix's DMs on IRC :>
narrow verge
EsperNet!
soft hornet
ESPERNET!
modest bisonBOT
π
narrow verge
The best
modest forge
E S P E R N E T!
pallid yoke
looking fresh https://i.imgur.com/N0qLbOV.png
modest bisonBOT
I don't like that the data is not taken all at the same time
modest bisonBOT
I have no Idea how hangar works, and quite honestly I do not want to know as long as it just works
pallid yoke
π
main lava
How are you going about the upcoming changes to ore?
In a change of policy, the Ore review system will be changed from the current approval-based system, to the Community Reporting System, whereby we will provide a mechanism for the community to flag content that may be malicious and/or violate the TOS. As part of this, we will be removing the click-through warning that currently exists on unverified downloads and replace it with a banner disclaimer.
Importantly, this change means that our community will have even more input on these matters and we hope that you will help others by flagging that may cause harm. The exact design of this system is still under consideration, but we will keep you informed as it develops.
In addition, development on Ore API v2 is proceeding, with a focus on ensuring stability of the new system.
well, as usual, we have not agreed (or even really talked about) any policies, so we don't know if we wanna mandate approval before upload or react after a report
we don't want to be in a situation where we have large review queues, thats for sure
stuff like, only review first upload, then rely on reports, is also possible
main lava
Still It would be good to have a system in place to at least somewhat prevent bad jars.
pallid yoke
reporting is already implemented
main lava
Hm
pallid yoke
and yes, I kinda want to do a bit automated scanning
pallid yoke
that too
I want to flag all obfuscated resources and put up a warning on download
because its inherently unsafe for users
main lava
Yeah I saw a very obfuscated one on spigot yesterday
modest bisonBOT
Maybe also detect any uses of force-op and changing other files
main lava
They made methods not readable for the decompiler
And I had to read bytecode
Maybe detect that stuff and encrypted strings
pallid yoke
I wanted to look into using https://github.com/OpticFusion1/MCAntiMalware
and if that sucks too much, I would roll my own one ppl could contribute to
pallid yoke
it would also be extended when new reports come in
but, tbh, that is most likely out of scope for MVP
we need to get out the basic functionality in the hands of users first
main lava
For now flagging everything that can't be 100% decompiled would make sense.
pallid yoke
and then we can look into convince stuff like t
main lava
Hm ok
pallid yoke
I mean, doesn't mean we don't accept PRs for that rn, but I am personally looking to finish the UI first π
modest bisonBOT
From my experience krakatau appears to be a good decompiler in that it's output is easily parseable, only issue is that it takes far longer (like 60x longer) to decompile anything compared to ForgeFlower and that the full classpath needs to be known beforehand in order for it to decompile fully
pallid yoke
no need to decompile tho
modest bisonBOT
yeah
ivory ingot
More progress when button alignment is done. π
pallid yoke
which buttons?
ivory ingot
Generate, Upload, Rename.
modest forge
Ctrl-F "setOp", done
ivory ingot
s/setOp/setBanned/ would be fun as well.
soft hornet
Ctrl-F "setOp", done
Also detect classloader stuff!
modest forge
Nah, on Hangar we're all living on the edge π
ivory ingot
And detect the java version and add a tag that the plugin is only compat with Java x and later.
pallid yoke
hangar will require plugins to be compiled with java 17, clearly
modest bisonBOT
Ouch, that means I have to get a 4th JVM
pallid yoke
why keep old versions?
modest bisonBOT
Modding non-minecraft java games
pallid yoke
thats not an explanation, as its backwards compatible?
modest bisonBOT
Well, libGDX has some issues when it comes to OpenJDK
Well, it's actually ld.so that is the one that creates the issue, but idk what really happens there
main lava
And detect the java version and add a tag that the plugin is only compat with Ja...
No. Just 
main lava
hangar will require plugins to be compiled with java 17, clearly
And also require latest snapshot 
gritty iron
Found it
modest bisonBOT
modest bisonBOT
I see (or not, 'eh)
'eh? what are you one of those syrup drinkers now?
I kinda learned English from Canadians π
oh no
pallid yoke
Worth noteing that that is the old branch
But https://hangar-new.minidigger.me is bork iirc
Ye
fluid oyster
Pls unbork
pallid yoke
for some reason it seems like sass is broken?
pallid yoke
yeah I have really no idea
ERROR Invalid or unexpected token
@import './tools/_index'
^
SyntaxError: Invalid or unexpected token
at new Script (vm.js:101:7)
at createScript (vm.js:262:10)
at Object.runInThisContext (vm.js:310:10)
at wrapSafe (internal/modules/cjs/loader.js:967:15)
at Module._compile (internal/modules/cjs/loader.js:1027:27)
at Module._extensions..js (internal/modules/cjs/loader.js:1092:10)
at Object.require.extensions.<computed> [as .js] (node_modules/ts-node/src/index.ts:1045:43)
at Module.load (internal/modules/cjs/loader.js:928:32)
at Function.Module._load (internal/modules/cjs/loader.js:769:14)
at Module.require (internal/modules/cjs/loader.js:952:19)
nuxt docs say "In case you want to use sass make sure that you have installed sass and sass-loader packages. "
so I did that
no change
paper heart
what did you change?
also - i'll take a dig at approval versioning (pages/admin/versions.vue) in a few hours 
pallid yoke
issue is, I dont remember the last time it did work
since it only breaks on yarn build && yarn start, not on yarn dev
and like, yarn start isnt throwing the error, only after the first request comes
and I dont keep logs of dead containers
so no way to tell when it last worked π
could be weeks
paper heart
i'll give it a shot on my machine see if it builds properly
yeah builds fine for me
maybe try clearing your node_modules @pallid yoke?
oh wait
pallid yoke
do yarn start and do a request
paper heart
no it doesnt π
package.json doesnt seem to contain sass or sass-loader
maybe manually adding those would get us working?
that just broke vuetify even more
pallid yoke
I have that locally
paper heart
what ver?
pallid yoke
paper heart
mind jest doesnt seem to be in the new frontend
π
the stacktrace is really useful
random badger
I may have added that to make it shut up about peer dependencies
pallid yoke
I know that I downgraded it to 4 from 5.16
paper heart
seems to be quite a niche issue though, as any similar issue i could find seems to be related to jest which isnt being used on the new nuxt project
pallid yoke
at least thats what git blame says
paper heart
Is there a list of phrases, or a way to generate phrases?
ah, clocked the typescript file
mellow pebble
Twitter
modest bisonBOT
@Phil: old frontend https://github.com/PaperMC/Hangar/blob/master/src/main/frontend/src/components/entrypoints/admin/VersionQueue.vue
underReview and notStarted are List<ReviewQueueEntry>
so just create a new interface for that class somewhere and hardcode some dummy values in the page
pallid yoke
@paper heart
paper heart
cheers
deft drift
ERROR Failed to compile with 1 error 1:39:11 AM
This dependency was not found:
* @johmun/vue-tags-input/vue-tags-input/publish in ./node_modules/cache-loader/dist/cjs.js??ref--10-0!./node_modules/babel-loader/lib!./node_modules/cache-loader/dist/cjs.js??ref--0-0!./node_modules/vue-loader-v16/dist??ref--0-1!./src/components/entrypoints/projects/ProjectSettings.vue?vue&type=script&lang=js
To install it, you can run: npm install --save @johmun/vue-tags-input/vue-tags-input/publish
Webpack Bundle Analyzer is started at http://127.0.0.1:8888
Use Ctrl+C to close it
^C
β― yarn add @johmun/vue-tags-input/vue-tags-input/publish
yarn add v1.22.10
[1/4] Resolving packages...
error Couldn't find package "@johmun/vue-tags-input/vue-tags-input/publish" on the "npm" registry.
info Visit https://yarnpkg.com/en/docs/cli/add for documentation about this command.
also, trying to start hangar api itself https://paste.gg/p/anonymous/2a0dbb64e1154314a9351db538136bc2
and yes el DB is running https://static.jacobandersen.dev/shared/ππ½β½οΈπβπ.png
deft drift
ok cat informed me that the active branch is nuxt-frontend
so im trying that now
k seems to load except for the avatar bork, idk if that's a setup issue or bug
Error occurred while trying to proxy request /avatar/paper?size=120x120 from localhost:3000 to http://localhost:8000 (ECONNREFUSED)
sorry irc, lmao
modest bisonBOT
?
deft drift
sent 2 msgs by accident
modest bisonBOT
No real issue
deft drift
where is this channel linked, not on esper with the rest?
modest bisonBOT
Correct.
there
didnt have the -dev prefix
suffix
fuck im out of it tonight
leaden lily
RE: the latest commit. If you set the fallback to an external source, won't it have to reach out to it for each one that needs a fallback? Couldn't you just inline it into the static directory or something?
pallid yoke
thats not prod code, there is never a fallback necessary as hangarauth handles that, but am too lazy to start hangar auth locally
also, no, your browser is smart and caches stuff
pallid yoke
yeah thats not why I added it
and like, you will hopefully still have most images cached, or cloudflare will
random badger
oh mini, did you figure out that spring security issue?
random badger
oh ya ok. maybe a better solution would be to have a getGlobalPermissions method that returned the default globals if no principal existed?
I feel like that might be useful other places as well
pallid yoke
ah yeah
random badger
hmm Mini, I'm thinking we just need to add a separate method for checking each permission since we can't use typescript enums the vue templates themselves
making a bunch of redundant get canEdit, get canDelete which is kinda annoying
would be very nice if we could do $util.hasPerms(NamedPermission.DELETE_VERSION) inside a v-if but I don't think that'll work
pallid yoke
Yeah main reason i didn't implement all the perm checks is that u couldn't figure out a nice way
random badger
Idk how to fix this issue https://github.com/PaperMC/Hangar/commit/7af8e89398f52e44251abf15da391529c83e7c99#diff-f08885ca2e7e78828de8b9714ee08c31a8a938bc3137f6e048f82024bc06efaaR14
random badger
Took me a solid 15 minutes to realize/remember that it was broken for me when I tried to use hangar yesterday
pallid yoke
Am sorry :/
sly mason
modest bisonBOT
Putting that on full Volume wasn't the wisest of all ideas
pallid yoke
Gotta post the motivational shia before i go to bed, smh
deft drift
.remind -d -w 126975485493248000 13h 819330968883953704
idk its bork
also not sure what happened to that link but ok discord
errant jacinth
hmm Mini, I'm thinking we just need to add a separate method for checking each p...
TS enums as a concept are generally considered an anti pattern. A type union of strings is generally the way to go
Also means it works properly with vue etc
random badger
Yeah, I only realized you could do that yesterday. Way nicer for some routes I think.
random badger
damn... I love Vue
random badger
hmm Mini do you remember how this review thing is supposed to work?
like can you add a message here now on your admin acct?
https://hangar.minidigger.me/MagicCheese1/DamageIndicator/versions/1.0.1.10/reviews
I see this
pallid yoke
But basically, i expect the list down there to have stuff like started review, stopped, approved, approved partial, and "user added a message: something here seems fishy"
Like, notes
Message might be a dum teen
Term
random badger
I think the way its setup, is that only the person who started the review can comment until they stop or approve
pallid yoke
I guess that makes sense, don't want ppl to duplicate work
So like, I start review, I see something I don't understand, I add message, I stop review, it goes back to the queue again and somebody else can look and sees my note
random badger
hmm ok. sounds good
pallid yoke
I don't exactly know how ore does it, but that seems like a sane flow to me
random badger
btw, I tacked the version dependency stuff finally. I had kinda been dreading do all that, but it should be all working, including editing platform versions and plugin dependencies after the version has already been created.
oh, do we wanna mess with the JSONB column of the project version review table?
basically as far as I can tell, each review is one row so all the messages from the user and stored in the JSONB column
could add a new table, project_version_review_messages table
pallid yoke
I really don't mind much, whatever works best I guess. A table might be easier to work with?
And yeah, I did fly thru the commit earlier, saw the perm stuff, gotta remember that that's now easy so i don't need to skip it anymore ^^
random badger
table is def easier to work with
pallid yoke
That's what I thought, let's just go with that then
sly thunder
Is there a roadmap somewhere of stuff that needs to be fixed?
pallid yoke
Most stuff is handled with Todo comments
The easiest to pick up are todos like "implement XYZPage", there we need the whole layout
sly thunder
gotcha
sly thunder
im probably being stupid again but i cant get the postgres server running i'm in the docker/db folder but when i run docker-compose up its complaining about hangar auth? I thought i didnt need that?
hmm doesnt work from the docker folder using the root docker-compose.yml either
ah yes, smoll brain strikes again, got it working now
pallid yoke
I only run the dev db compose file, you can tell it to use a special file with -f
But IJ also has clicky stuff to make it wasy
sly thunder
yeah thats how i got it working
well it works, but its not having a good time
its not loading any resources
pallid yoke
Have you yarn run dev or whatever?
You need db, spring and fronted
I do db in docker and rest in IJ
I think MM does everything in docker
pallid yoke
Remember, frontend here also renders all html on the server
sly thunder
render as in building the page and sending it to the client you mean?
Failed to compile.
./src/components/entrypoints/projects/ProjectSettings.vue?vue&type=script&lang=js (./node_modules/cache-loader/dist/cjs.js??ref--10-0!./node_modules/babel-loader/lib!./node_modules/cache-loader/dist/cjs.js??ref--0-0!./node_modules/vue-loader-v16/dist??ref--0-1!./src/components/entrypoints/projects/ProjectSettings.vue?vue&type=script&lang=js)
Module not found: Error: Can't resolve '@johmun/vue-tags-input/vue-tags-input/publish' in 'C:\Users\Joshua\Documents\Projects\Hangar\src\main\frontend\src\components\entrypoints\projects'
i ran yarn install and then yarn run serve
sly thunder
ok
sly thunder
ah
pallid yoke
Use nuxt branch, frontend is now in root
sly thunder
hm, how do i get that to run? yarn run serve doesnt seem to work in Hangar/frontend
install worked fine tho
pallid yoke
So am I ^^
sly thunder
god this takes so long to build
https://pastie.io/meuyva.md well, now i'm getting migration errors
sly thunder
if i delete the docker container all data should be removed right?
woven ridge
if you didn't mount anything, yes
sly thunder
idk man, im just pressing green play buttons
woven ridge
don't really remember how desktop has that set up
woven ridge
sounds like it
but on the other screen there is a section for volumes/mounts, if nothing is there than yes
sly thunder
it just comes back
Β―_(γ)_/Β―
and when i try to run it i get: Failed to deploy 'Compose: docker': Sorry but parent: com.intellij.execution.impl.ConsoleViewImpl[,0,0,1275x227,invalid,layout=java.awt.BorderLayout,alignmentX=0.0,alignmentY=0.0,border=,flags=9,maximumSize=,minimumSize=,preferredSize=] has already been disposed (see the cause for stacktrace) so the child: com.intellij.util.Alarm@4a213a01 will never be disposed
woven ridge
oh i got no idea about how intellij works with docker
pallid yoke
Inteloij docker is still sketch
Docker stores data in volumes
So if you wanna nuke data, gotta nuke that
Or login and delete the schema
woven ridge
if you just want to start fresh you can just do
docker stop $(docker ps -a -q) && docker rm $(docker ps -a -q) && docker volume rm $(docker volume ls -q)
or something like that, haven't tried it
sly thunder
woven ridge
windows docker might be different somehow Β―_(γ)_/Β―
because that's for sure a thing
just shows stopped containers
sly thunder
... i deleted the container, removed the volume and its still bitching about migrations
pallid yoke
Well, use IJ to login into the db and see
Maybe jake broke the migration yesterday π€
sly thunder
i managed to get it deleted
straight to the next error
[Template Error] The following has evaluated to null or missing: ==> config.useWebpack [in template "utils/hangar.ftlh" at line 6, column 10]
pallid yoke
Why is that still freemarker π€
You sure you on nuxt branch?
Oh most likely you accessed it from the wrong port
Try the port nuxt tells you in frontend CLI
sly thunder
well thats what it says at least
sly thunder
yeah makes sense, it didnt work earlier on 3000 cuz the spring app wasnt running yet
seems to work now
well that was a process...
pallid yoke
Isn't technology fun? π
If you want, i guess a good way to contribute, and to make it easier for the next ones, is to write what you just learned down. For me and Jake that's hard, because we are so used to all the quirks, from this projects and others, so it's hard to notice when stuff isn't obvious
sly thunder
yeah i suppose i could do that
should i just put the entire process on the root readme?
pallid yoke
If that's where you would expect it, sure!
It it's too long, might be worth expanding it into a contributing.md, but always can do that later
I guess the readme in frontend is usless anyways, can just nuke that
sly thunder
yeah its a bit on an odd place
pallid yoke
It's auto generated
sly thunder
oh
@pallid yoke so what would be the proper docker command to run the db? (i just pressed the play button)
sly thunder
what does the -d stand for?
pallid yoke
Start in background iirc
sly thunder
ah
pallid yoke
Demon
sly thunder
this is what i have so far
pallid yoke
Sounds good!
random badger
Ded, did you get it working in the end?
sly thunder
Ded, did you get it working in the end?
yeah
sly thunder
alright, will do
pallid yoke
I can't review tonight, can get to it tomorrow if nobody else merged till then
sly thunder
Sure. It's not like it's a big deal hah
random badger
for version reviews (this is for moderation of versions) would you expect the oldest review to be at the top or bottom of the list?
random badger
so right now, the newest is at the bottom.
there is an option to hide stopped reviews so only the latest unfinished one is shown
obtuse orchid
most recent first makes sense to me
random badger
but then it might get kinda long where you have to scroll all the way down to get to the newest stuff
random badger
reviews for a version
narrow verge
like approvals?
random badger
yes
obtuse orchid
oh well if its a thread then prob the other way around
i thought you meant like user reviews for a specific version
narrow verge
Oldest unresolved should be top
random badger
no no, not user reviews
narrow verge
dont think were doing user reviews even
random badger
no. well there's a forum thread attached to your project if you want it
and people can talk about it there
obtuse orchid
ya idk im just randomly stopping by this channel lol
random badger
k, so I'm changing the functionality of the version approval system a bit
before, only 1 review could be ongoing by a moderator
now, each moderator can start their own review and either stop, approve, or partially approve
then they can go back and re-open their own review.
no more "takeover" the review from someone else cause you can just start your own at the same time
narrow verge
no sniping reviews?
pallid yoke
The most important thing is others being able to see the messages i guess
I don't particularly care about the order, in 99% of cases the list will not have more than like 5 entries anyways
random badger
Itβs possible I spent entirely too long on the review page...
pallid yoke
its possible be spend too much time on a lot of things π
but it will be all worth in the end
random badger
I think it looks pretty π₯
Reactivity just makes it soooo easy to have stuff just work
pallid yoke
oh review is looking hella fancy
pallid yoke
oh man am too dum for sql
SQL method returns primitive long, but statement returned no results
this makes no sense
ah got it
random badger
oh review is looking hella fancy
Ikr?
pallid yoke
mmmh the perm util doesnt seem to work for project perms
:can-edit="$perms.canManageSubjectMembers" results in false
even tho am owner in that project
I like how almost everything on project page is down, but the download link π
cause who needs that
random badger
Hmm. Is it checking the right permission?
pallid yoke
found it
I had canEdit: Boolean = false;
cause I thought that would work
as a default
but vue sets the prop and the false then mutates it again
pallid yoke
does async data work on layouts?
random badger
hmm. I guess its a "page"
pallid yoke
it doesnt seem t work
random badger
oh, you should probably use the nuxtServerInit store action instead
that's where a lot of global stuff is fetched
pallid yoke
fetch works
random badger
what are you getting in layout?
pallid yoke
announcement
random badger
hmm, fetch isnt blocking when loading from client-side tho, that gonna be an issue?
pallid yoke
well, am trying at least
clearly spring thinks otherwise
lets see
ok its called both on client and server?
I dont get it, lol
on page load I can see it in html, then it vanishes
ok now it stays
and its only executed on the server
I just looked wrong, nuxt sends server log to client too
random badger
yes, that's relatively new. threw me off when I first saw it too
I was like.. why is it logging in both places
pallid yoke
its easier to have everything in one place
and its marked as SSR log
I like it
just need to look properly π
I only partially approved the project
thats worth to show, right?
heh to approve full after partial, you gotta undo and then approve again
anyways, works
I kinda also want to have push notifications
but that will have to wait, lol
heh, going thru old issues
"I mostly was speaking about colors, I don't think its practical to switch away from bootstrap, we use quite a few of its features"
haha
random badger
I have no clue how push notifications work on a web app
random badger
"I mostly was speaking about colors, I don't think its practical to switch away ...
LOL
pallid yoke
neither do I but I know where to look
gotta use a service worker for that iirc and stuff
my irc client does that, I looked at the code before
random badger
oo, just realized, cause there can be multiple reviews, should it only notify users that the version has been reviewed if a review was approved while no others are open?
pallid yoke
Re: the stargazers stuff, should we just move that to internal then? Is there a use in having that public?
random badger
I think it originally was public api
pallid yoke
Mmmh
I didn't do that join because the query was already so big
But meh i guess if these queries become an issue in the future we can just cache both in spring and in CloudFlare
random badger
Itβs not a big thing to add it right? Itβs just a 1 line subquery?
pallid yoke
oo, just realized, cause there can be multiple reviews, should it only notify us...
Well, there can only be ever one result, right? So like, if maybe if I review and another review is open, warn me if I really wanna close? Idk
pallid yoke
Itβs not a big thing to add it right? Itβs just a 1 line subquery?
Yeah
Lmao dependabot
I changed the default branch and now it's not happy
Guess we can exclude the frontwnd old
pallid yoke
kek IJ showed me that the prop was dum https://i.imgur.com/BsSBGbI.png
oh I see why deployment is failing, I did a docker update or smth and it no longer exposes the socket, lol
thats rather sad cause that means I need to restart docker which means downtime
urg
gotta do that tonight I guess
but doesnt matter since nuxt start still is broken
pallid yoke
ok, I tracked down a nuxt discord, time to see if they know what is wrong
random badger
Iβve tried asking for help in the nuxt discord... itβs never done anything for me.
Mainly, itβs a lot of people asking questions it seems.
pallid yoke
yeeee
I am facing some weird issues with a query you wrote for getUserWatchingCount
the query looks fine
I get this Undefined attribute for token '<endif>' [statement:"null", arguments:{positional:{0:paper}, named:{user:paper}, finder:[]}]
random badger
The one already committed?
random badger
Oh, itβs missing the @UseStringTemlate annotation thing
When you use the <if()> stuff, you need that annotation on the query
pallid yoke
uufff
thanks didnt see that
its late...
hwo was that named route thingy thing again?
random badger
Itβs just the folder/file name separated by hyphens
So like βauthor-slug-versionsβ
pallid yoke
do you remmeber where you used that?
random badger
Hmm not off the top of my head. Search the project for βauthor-slug-versionsβ
pallid yoke
:to="{ name: author-slug-versions-version-platform, params: { ...$route.params, platform: version.toLowerCase() } }"
ah
random badger
Yep
We can custom name routes if we wanted to make that easier...
Right now nuxt just generates them automatically
pallid yoke
wonderful
I implemented the api keys as a page again
I think its too big for a modal
random badger
Hmm ya ok
pallid yoke
unless you got an idea on how to compact that?
I thought about making two, for new and for managing existing
but its easier this way
enough for today
slowly but surely we are getting stuff done
random badger
Well you can probably make those checkboxes shorter in height. Add the βdenseβ and βhide-detailsβ props to v-checkbox
pallid yoke
kek I tried to make stuff denser but didnt see props I think
so I wrote css
move the text field top because that makes more sense
.v-input--checkbox {
margin-top: 0;
}
is still needed tho
random badger
Yeah, I know I still added some css after the props
random badger
Speaking of checking boxes on a version... should probably change up the version table to have a bitset of βflagsβ for stuff like that.
That way we donβt have to keep changing the db table in the future
A couple of the Booleans on that table can be converted to that as well
pallid yoke
Yeah i had for usecases for that, cause they also should be used for filtering, be displayed as badges etc
random badger
Yep
pallid yoke
Opensource, obfuscated, "reputable"?, Antivirus flagged?
random badger
Well, open source might be more project-level
pallid yoke
Same with reputable
random badger
Same w/reputable. But yeah, can have flags on project as wel
For open source, should we have a preset list of hosts (GitHub, bitbucket, etc) that give your project the open source badge?
Idk what a good way to verify that would be
pallid yoke
Meh, we can just make that work with the source link, if you enter smth you get the badge, if you get caught cheating you get the boot
Cause a few ppl use their own git hosting
Oh and we have the licence thing anyways
If it's MIT it's open source, no?
random badger
Mm ya
misty mountain
https://spdx.org/licenses/ check their licence against this ;)
you'll have to take the decision whether FSF, OSI, or either are most important (which I know will be controversial among certain people), but otherwise a good list; other large projects like Rust's crates.io use it too
misty mountain
as long as that means jroy can use his beloved Giant Penis License 
modest forge
π
pallid yoke
oh shit proxi discovered our bad hangar code
modest bisonBOT
no good
misty mountain
A lot better than what I deal with daily 
though I do have a longer monitor at work, so some of this wouldn't go noticed :p
pallid yoke
well, thanks for looking at it
the more eyes on the code the better
I dont claim to know wtf am doing here π
pallid yoke
lol where
misty mountain
VersionApiController L48
pallid yoke
oh thats generated code
and in controller old
all packages with old in the name are deprecated
misty mountain
wai it exist
misty mountain
ic
pallid yoke
we trying to get rid of all of that with the nuxt rewrite
pallid yoke
flash was the term ore uses for alerts I thinkl
misty mountain
I see
pallid yoke
if you wanna do some real work, you can look at getUserWatching/getUserStarred and implement a getUserOrganizations, that returns a list of the orgs that user has π
pallid yoke
thats a todo in frontend
AuthorPage line 74
the sql query kinda shoudl exist somewhere in the old code too
SELECT o.id, o.created_at, o.name, o.owner_id, o.user_id FROM organization_members om JOIN organizations o ON om.organization_id = o.id WHERE om.user_id = :id
altho you prolly wanna join users so you can do where user.name
random badger
Erm... donβt really need to make changes to stuff in any package with βoldβ in it right?
That stuff isnβt used anywhere (or shouldnβt be)
pallid yoke
yeah that why I explained
random badger
Oh ok. I just looked at the commits
misty mountain
π delete old code
pallid yoke
sorry proxi that was wasted effort
misty mountain
wtf wtf wtf
random badger
Yeah, basically, donβt touch anything with βoldβ it itβs package name
pallid yoke
but until then its not worth deleting small stuff, and we can easier use it for reference
mini does constants in UpperPascalCase
pallid yoke
I cannot guarantee that is wasnt drunk when writing parts or everything of hangar
but I do have sonar lint, that yells at that normally?
misty mountain
no, it doesn't
misty mountain
we moved from qube to cloud :p
barren shale
the bbcodeconverter still in an old package smh 
misty mountain
good .
modest bisonBOT
but the bbcodes!
random badger
Ya, itβll prob get moved over. Unless proxi decides to rewrite it or smth... π
pallid yoke
its also not implemented in the new frontend yet
thats when we would mvoe it over I guess
pallid yoke
it even has a small thingy in the bottom right of a page if a hot reload is happening
spiral current
so the commit after that broke it?
pallid yoke
thats not too old
random badger
and... found a broken one not too long after https://github.com/PaperMC/Hangar/tree/e4e5644d028e01312ff792a756b588f9a2689efe
pallid yoke
~15 commits in between
random badger
git bisect... here we come
spiral current
random badger
is that it?
pallid yoke
its compare between the two
spiral current
yep
random badger
ya, could do it that way too
pallid yoke
css: ['~/assets/main.scss'], is new, together with main.scss
but has no vuetify at that point
random badger
are you sure its vuetify?
pallid yoke
the code is from vuetify
random badger
I didnt see anything about it in the stack trace of the error
pallid yoke
I looked at it before
its scss code from vuetify and smth is reading it as js
while you guys look at that, lemme restart docker so we can deploy again, lmao
urg ppl online on my teamspeaks
random badger
narrowing in on the issue
pallid yoke
maybe I should just rewrite my stuff to not use an exposed docker host anymore but just ssh
random badger
good: 4ca8d85b7254c146de266e0bdb1302c35189b821
bad: 2f936baf6bcbf5c47b7332c8f779f70814171375
spiral current
moment
pallid yoke
spiral current
ah, thanks
random badger
oh, that's gonna be it
pallid yoke
trying
random badger
that has smth to do with vuetify
leaden lily
Might have found something.
random badger
yeah, https://github.com/PaperMC/Hangar/commit/dd0c115fd4b92ce1229fb04104f8989dd25a3996 is the bad commit
just finished the binary search
random badger
did it work if you didnt move it out into its own plugin file?
I guess it did, cause that was the commit that stopped it working
I don't understand why that would be... but π€·ββοΈ
pallid yoke
so we cant use vuetify stuff in a plugin
leaden lily
That's so weird.
random badger
its useful code for sure I think, to dynamically change background color
leaden lily
Is it safe to push that commit so I can pull that?
spiral current
a question - i suppose I need to have hangarauth running in order to log in as the fake user?
pallid yoke
no
press login and refresh manually
ill commit my stubs for now and whenever somebody has time they can look at it
leaden lily
Oop. Just did a package update and now it won't compile anymore. Apparently we aren't on the latest version on Nuxt atm that had a set of changes so we'd have to modify our shit to make it compatible again.
random badger
yeah, I had that happen when I tried to update nuxt as well
will have to look into that
should look into that before too long tho, useful to stay on latest version as long as its not changing to vue 3. probably wont update to that for a while. vuetify for vue 3 just went into alpha, so itll be a while before all our stuff is compatible to update to vue 3
leaden lily
It looks to stem from the nuxt-il8n package. Looking into that right now.
random badger
well, there's something that the latest nuxt version has typescript compilation bundled with it now as well, so no need for the nuxt-ts command as well
leaden lily
Yeah that too.
pallid yoke
I remember having issues with that lib before
spiral current
btw fix is verified workingβ’οΈ
L143 in frontend\pages\notifications.vue has a spelling error - should be "Couldn't update invite"
random badger
well... it really should be not there
random badger
notifications was like the first page we did after home I think, prob before i18n
spiral current
well, it is used
pallid yoke
oh its just error log
spiral current
pallid yoke
oh its displayed to user
spiral current
yep
leaden lily
Okay so I've tracked down that new error on the updated stuff and it's just upset about how we and creating our config option. Working on making a fix.
pallid yoke
yeah that needs i18n
random badger
well, itll have it on the server end I think
I just replaced it with this.$util.handleRequestError
leaden lily
Oh Machine I wonder if you have to update this. It looks like it's directing to your fork of it fixing a memory leak?
pallid yoke
thats where I stopped too last time π
leaden lily
The error is stemming from vueI18nLoader: true
If you set it to false, it works no errors.
random badger
oh ok. ya my fix isnt the right fix according to the author https://github.com/nuxt-community/i18n-module/pull/1040
but apparently the right fix requires some big changes to nuxt itself
the mem leak issue isn't a problem for production, only for dev with reloading on changes
leaden lily
Okay so then I should be good to change this to use the real one?
random badger
I don't think its been fixed? maybe I can just rebase my fix on the latest nuxt i18n version?
leaden lily
That would be appreciated if you could do that so I could test this
random badger
mmhmm, will do
leaden lily
Thanks. Lmk when that's deployed so I can pull it and test.
random badger
ok, branch is updated @leaden lily
leaden lily
Alrighty. Let me try to upgrade that package.
random badger
we can just copy the code vuetify uses for those functions right? vuetify is MIT licensed
leaden lily
Hmm, I'm getting 504 errors when I try to view the page via yarn build.
But, I did get it to compile with updated dependencies.
random badger
probably because you need to have the server backend running
leaden lily
Well Mini said I could get it to run without a backend 
random badger
I mean... if you get it to compile, you can prob just pr the updates and be fine
no major breaking changes anywhere afaik, was just complicated because nuxt-i18n is pointing to my branch
should probably figure out a way to have some page show up without the server running
leaden lily
The only thing I don't think I can test is the removal of the typescript-runtime package because I can't load pages π
Well, lemme try.
random badger
k, well error page will show now
was just missing catching an api request on the home page
spiral current
are project approvals implemented on the backend?
random badger
version approvals are. I don't think project ones are yet
spiral current
i could try to impl the frontend
random badger
that takes place when you try to change the visibility of the project on the settings page iirc
ik mini started the settings page, but idk how far he got
random badger
like I think the way it worked, was a mod/admin could change the project visibility and if they changed it to need-changes, you'd have to give a reason why
(doesnt have to work the same as before, to be clear)
spiral current
and version approvals are?
random badger
yes.
leaden lily
Alrighty. PR opened.
random badger
only 1 tiny part is
Glare, I get this when I try to run yarn dev
I installed packages after checkout out your branch
leaden lily
Hmm, I think I saw that too. Happened when I updated nuxt.
Cause I don't natively see that package in our package.json
random badger
@typescript-eslint is probably required by @nuxtjs/eslint-config-typescript
leaden lily
I could try downgrading typescript itself back to the version below 4.2.0?
4.1.3 looks to be it
random badger
yeah, hopefully a package doesn't require > 4.2.0 then
leaden lily
We'll find out
Oh I see why
"@nuxt/typescript-build@^2.0.4looks like it brought in 4.2
leaden lily
So what's your suggestion for that warning we're getting right now?
random badger
well it looks like that typescrpt-estree supports 4.3.0 now (in its repo at least) https://github.com/typescript-eslint/typescript-eslint/blob/0473674c58df5039a2de3c63ad7494fc6be7487e/packages/typescript-estree/src/parser.ts#L28
leaden lily
So then we should be good
pallid yoke
can you pin all deps?
random badger
what version of typescript-estree does it have?
wdym "pin"? remove the ^ before them?
leaden lily
"@typescript-eslint/typescript-estree@4.14.1
random badger
hmm ya that's outdated
pallid yoke
ye, and set them to the real version, because else its hella confusing
also
Updating service hangar_new_db (id: fibnjz43p5k4urly8fh53moof)
Updating service hangar_new_mail (id: fn3fgrcmwpm8ocdrbh1509kzp)
Updating service hangar_new_auth (id: yyva13y3wcgidhhwgwqxa9fhe)
Updating service hangar_new_redis (id: ed6449og9jt89ub5k5wzgxmna)
Updating service hangar_new_frontend (id: na4bfurxw9uudpqsh0hmvys1m)
Updating service hangar_new_backend (id: qa6re6tw2r5zmf9x05fdm35hx)
yey
now to nuke db I think, lemme chekc backend logs
ye
leaden lily
Lemme try to update that specific package.
Idk why it's not showing up on interactive.
random badger
ya idk either
they are required here I believe https://github.com/nuxt/eslint-config/blob/master/packages/eslint-config-typescript/package.json
random badger
.... smh
pallid yoke
rip old staging
random badger
hmm Glare, I just ran yarn upgrade, and it upgraded a bunch of stuff
does interactive not show everything?
random badger
interactive maybe only shows deps in package.json
pallid yoke
yes π
random badger
ugh... I was used to testing how stuff worked on the projects people had put on there
π
pallid yoke
yeah π
leaden lily
Okay I ran it without interactive. and not getting that error anymore.
random badger
yeah, it seems yarn upgrade upgrades everything in the .lock file where yarn upgrade-interactive only shows dependencies in the package.json
ya, it seems to work fine with all upgrade packages, so just add that. and then do what mini suggested and remove all the ^ from the package.json
leaden lily
I'll go ahead and push these changes.
Hmm could've sworn there was a command you could run to pin it all.
leaden lily
" If you are using yarn to manage your node dependencies, you will automatically have dependency pinning due to the yarn.lock file that yarn produces and uses. "