#Packaging solutions for Modrinth app (see pinned)
1 messages · Page 2 of 1
I don't think major distributions do (debian/ubuntu/arch), whereas webkit seems to be in everyone's repos
idk, I think webkit is prolly a better choice solely because of that^, even though it seems to be incredibly inconsistently built by distro maintainers
so you can have a pretty light weight package (see modrinth deb package, just 5 mb), and not ship a chromium every time
Not really. Microsoft's been shooting themselves in the foot and ass at the same time with the recent hellhole that is Windows 11, and its just gonna get worse and worse, and more and more people are gonna move to Linux as it does
Ik Prism can handle Modrinth packs too, but for a new user, having a CurseForge app-like UI sounds less confusing then MultiMC's UI imho. I only brought up Flatpak/Flathub due to it just, being easier, and being setup OOTB on most distros
And either way, if more & more people are moving to Linux due to Windows getting more sucky, that means more people will use the app on Linux. Just give it some time
I thought Tauri was Firefox
issues with windows 11 doesn’t make linux less of a niche. people also always say that every new release of windows will lead to more linux users. it doesn’t
prospector is completely in the right
and that’s fine tbh. desktop linux doesn’t need to be super popular, and either way we have an easy way to get the app now anywhere
Ik Linux won't be super famous, but its gonna get to a point where Windows is gonna become so bad that people should at least try to move, instead of suffering lol
No, Tauri uses WebKit2GTK on Linux
And Blink is forked from WebKit
Gecko is not
I would say that many windows users are switching because of windows 11. In the last 4 years linux has doubled its own market share. It previously took well over 20 years for 2% more. Many say the steam deck is responsible for this, but I don't think so because very few people use a browser with the steam deck. Many computers do not support Windows 11 and if windows 10 no longer gets support (70% of windows users are windows 10 users) many will switch to linux because especially many gamers have now realized that it works.
Linux mint uses flatpak by default but will now only show verified flatpaks in their software store in the future (I don't want to be a pain, but that would be one more reason to verify the flatpak. And by the way, I think "maintained by getchoo" would be a good idea)
you are forgetting steam deck users and people with touchscreens. I would say modrinth UI is much more touch friendly than Prism and others
hopefully at some point after 1.0 we can add proper controller support as well
Huh, didn't know
Controllable exists?
Whoops, true lol
#1138510665947041876 message
I agree, the Deck defo helped a lot.
What needs to happen now is Nvidia needs to fix Wayland & make it not a buggy hit or miss mess
And Discord fixing screensharing/audio
and for desktop linux to become less of a hit or miss mess
True lol.
|| hit or miss. I guess they never miss, HUH? you got a boyfriend I bet he doesn't kiss YA! … ||
Now that the flatpak is ready, would it be a good idea to add this to the app's download page?
I can make a PR for it, the link goes to the issue which has all of the information for each package
Alternatively, it could be this, if you only want to show the Flatpak
we should probably make a support article about the third party packages
and link to that
And probably mention that the app is now on Flathub in an announcement later on
so should i set it as "Download third party packages" and then once the article is made I can PR
or maybe just "Third party packages" to keep it short
yeah probably just third party packages
maybe make a community distributions section akin to download one and list third party distribution options there like winget, flatpak, etc
yeah that works
are support articles available in a git repo, like can i PR them? or do authorized users have to make them
it's fine if i cant i just kinda wanna help out somehow lol
nah only authorized users can but if you wanna write up some text for it, that could help
does Intercom use HTML or Markdown? or both?
anyways
neither
Collection: Modrinth App
Title: Third-party packages
Summary: How to install Flatpak, WinGet, and more
Written by: Devin (you don't have to put my name of course, if that's even possible to do)
While the Modrinth App is currently officially distributed from the app page on our website, there are also community-maintained distributions of our app for various different package managers such as Flatpak and WinGet.
Please keep in mind that these packages are not officially maintained by Modrinth staff.
Flatpak
If you are on Linux and want to install the Modrinth App using Flatpak, first ensure that you have Flatpak installed with the Flathub repository enabled. Then, you can use the software center if available, or alternatively run this command in your terminal to install:
flatpak install flathub com.modrinth.ModrinthApp
WinGet
If you are on Windows and want to install the Modrinth App using WinGet, make sure you have WinGet properly set up and then run this command in your terminal to install:
winget install -e Modrinth.ModrinthApp
Other packages
Interested in other packaging options? This GitHub issue is consistently updated with the latest packaging information from the community. Packages for Arch and Nix are listed here, among others.
oh
it's WYSIWYG
i see
Awesome! Thanks!
once the article is added lmk so i can make the PR for the app page :]
https://flathub.org/badges there are also badges
https://support.modrinth.com/en/articles/9298760-third-party-packages does that look good?
How do I install Modrinth App using Flatpak, WinGet or other packaging solutions?
changed the description to be the question for consistency
Mind if I ask why the flatpak is not a first party package ?
It is the closest we have come to be fully cross distro compatible and tbf is a much better target than an AppImage
tauri build appimage = ez we use dat
tauri not build flatpak = wth how it work
if that is the only reason, I guess that is fine
I hope to see first party flatpak support once it is easy to do it (when tauri adds easy support ig)
another reason is that I think none of the dev team are linux users lol
geo uses mac, pros uses.. windows? mac?. I'm no dev but also use windows
so it's just better for us to leave it to the people who know how it all done correctly, the linux way™
that is valid. cant support fully if you dont use it
i use windows yea
yup!!
will get the pr out soon, just not at my pc rn
This notes the third-party packages available for the Modrinth App on it's page. Links to the article I wrote earlier https://support.modrinth.com/en/articles/9298760
@gleaming trench
epic thank you
no problem
oh also maybe it'd be a good idea to update this article with a link at the bottom to the third-party packages page, with a disclaimer that it's not officially maintained of course https://support.modrinth.com/en/articles/8797818
Why isn't the Modrinth App packaged for my distribution?
@honest cedar @cold dove Big thanks for getting this whole thing set up, glad Linux/Steam Deck users can enjoy the app now too
yupp!!!
just waiting for this PR to get merged now
ok
PR?
thamks :)
pull request
we’ve been merged into nixpkgs!
https://github.com/NixOS/nixpkgs/pull/289149
i ended up getting mac support for it as well. idk if i ever mentioned that
🥳
Barely missed branch off before 24.05 though , didn't it? Or is it included this stable?
i am running into a issue where minecraft does not recognize resource packs where the name includes a § symbol for styling example resource pack
this does not happen when i run the dev version of thesues
I am not shure if the problem is related to the flatpak or the build process itself.
the problem is that i cant start the appimage it self since it doesnt start (driver issues.... )
would somebody be so kind and test if this also happens in the appimage
Thanks!
and big thanks for all the work done to create a flatpak 🙂
@cold dove Would be nice if the launcher opened on its taskbar icon instead of a Wayland icon thingy
@lost mesa Has no issue
Tested on 1.20, not sure about other versions
That’s probably just an issue with the launcher on Linux
Hence why I pinged Getchoo
they aren’t modrinth team tho
but i believe this happens because kwin on wayland (or maybe wayland in general?) finds the matching .desktop file with the same string as the app_id, which is this case would be like modrinth-app or something, however the flatpak’s file is called com.modrinth.ModrinthApp so it doesn’t show a proper icon. i’m not really sure if that can be fixed by the packages without having to use some weird hacky stuff
i think you can fix this with a kwin window rule
Ik, they do maintain the Flatpak though lol
not sure why it happens. the icons are recognized and associated with the app otherwise
might be because we use a launcher script?
but tbh i don’t know much about that stuff in flatpak
also i would prefer github issues for anything to do with the flatpak. it makes it much easier for me to track it with other things im working on
Ask the Flatpak devs?
when i have the time and can get in touch with one, yeah i’ll be right on that
Mk
Ngl, Flatpack really needs to get their stuff together and get Flathub to make sure apps have the needed perms set so everything just works (Discord and many others are examples). Flatpak's devs seem to be much more devs first, users last
the icon works just fine on GNOME
Tried installing it using nix, got this error.
error: flake 'github:getchoo/nix-exprs' does not provide attribute 'packages.x86_64-linux.modrinth-app', 'legacyPackages.x86_64-linux.modrinth-app' or 'modrinth-app'
It’s only a KDE issue, and the reason it happens is because of #1138510665947041876 message I’m pretty sure
doesn't gnome search icons in a non-compliant way
i guess it could be a gnome-specific fix idk
can't find but basically there was an issue for gnome that it loads icons differently from any other DE
https://bugs.kde.org/show_bug.cgi?id=459536#c2 found only this
sigh Of course they do it differently lmfao
They make things very annoying for everyone else
Tbh they should just drop out of all standard bodies they are part of, since if they have no interest in standards, they should get out of the way of everyone else trying to make an ecosystem like every other DE
This is why XApps from Mint became a thing
we’ve been merged into nixpkgs so i removed it from my repo. you can just use nixpkgs#modrinth-app now
ooh, i forgot to update the issue. give me a moment
yes (though nixpkgs can still technically be a flake)
it is on search.nixos.org, but only on unstable https://search.nixos.org/packages?channel=unstable&from=0&size=50&sort=relevance&type=packages&query=modrinth-app
Thanks.
shouldn't be a long wait if you're stable user either. we made it into the merge window for 24.05, so it'll be available for you guys in a few days when that releases
…lol
it makes sense why i didn’t run into any of these before trying out plasma 6
that’s now 2 bugs exclusive to it 🫠
search? no
name their own? sort of
Is there a way to install it now without unstable release or do I have to wait??
you could fetch unstable nixpkgs and use modrinth-app from it
What do you mean? I tried installing it now and got this error.
Wait, I can't send images here.;
[maksz@nixos:~]$ nix-shell -p modrinth-app
error:
… while calling the 'derivationStrict' builtin
at /builtin/derivation.nix:9:12: (source not available)
… while evaluating derivation 'shell'
whose name attribute is located at /nix/store/7d732w65sy3amw4nmyy3b3yk71qm6df4-nixos-23.11.7042.9a9960b98418/nixos/pkgs/stdenv/generic/make-derivation.nix:348:7
… while evaluating attribute 'buildInputs' of derivation 'shell'
at /nix/store/7d732w65sy3amw4nmyy3b3yk71qm6df4-nixos-23.11.7042.9a9960b98418/nixos/pkgs/stdenv/generic/make-derivation.nix:395:7:
394| depsHostHost = elemAt (elemAt dependencies 1) 0;
395| buildInputs = elemAt (elemAt dependencies 1) 1;
| ^
396| depsTargetTarget = elemAt (elemAt dependencies 2) 0;
error: undefined variable 'modrinth-app'
at «string»:1:107:
1| {...}@args: with import <nixpkgs> args; (pkgs.runCommandCC or pkgs.runCommand) "shell" { buildInputs = [ (modrinth-app) ]; } ""
| ^
[maksz@nixos:~]$ ```
that's because your <nixpkgs> is the 23.11 branch (or a non updated unstable)
try nix-shell -I nixpkgs=channel:nixos-unstable -p modrinth-app
or you can use the 24.05 beta with channel:nixos-24.05 instead of channel:nixos-unstable (though atm they're basically the same)
I said without unstable release.
unless you pull in the files from the unstable release, no you will need to wait
im just recommending using only modrinth-app from the unstable release without switching your whole system to it
whats the other one ?
apparently the window property of modrinth app says that the desktop file is called modrinth-app-wrapped which is kinda weird
yeah that’s what i figured would be the issue
i’m guessing kde doesn’t follow symlinks to get icons
i think i didn’t run into this during testing since when i originally made the flatpak, i just included my own desktop file since the auto generated one from tauri included in the appimage sucks
and i know i set Icon explicitly there. i dont think tauri’s does, though
(which is what i use now because a review told me to :/)
Really wish all these DEs used proper standards lol
Gnome does its own stuff, KDE does its own stuff, etc
me when the portals are still implemented by different developers with different features available, all with varying quality
the amount of bug reports i’ve seen exclusive to one implementation is crazy
Imo Linux and its components (Wayland, portals, etc.) need to be made so that they can only be implemented in a certain way
This would avoid the mess that is 90000 pointless distros, 500 package formats, etc.
Or... Apps that need permission for something could just ask the user to hit Allow or Deny, like on Android/iOS
This way things are handled automatically
this will never really happen as long as there are multiple implementers. you just need one thing
Ik, ugh
me when people contact support over aur package which we have nothing to do with demanding us to fix it .-.
this is precisely why I think approving flatpak would be a bad idea
I wish AUR also had a screaming UNVERIFIED label
damn end users... just use the hecking aur comments
AUR in general is supposed to be unofficial as it guarantees no support
also there is nothing wrong with limiting support for Linux. Even spotify does it
of course, but do you think the average user knows or even cares
a person using AUR should know and care
otherwise it is literally against how arch systems are maintained/used and they risk breaking their systems entirely and has various other security risks
arch users do not care about this
The AUR shouldn't exist imo. You're just asking to get infected with malware, lol
it's also just a bad platform to support
even if you do officially maintain packages there, you have no real control
a good example is with prism we "unofficially" support it, but all the maintainers are also maintainers of prism itself
we want to drop qt5 support at some point in the near future, but if we remove the package, anyone can just reupload one under the same name, become the maintainer, and systems will update to it
lol
And tbh, I really wish Flatpak would get better at handling permissions so things like Discord file drag/drop wouldn't need user intervention, etc.
the best way to handle those permissions would be to not have them
by default, anyways
How would file drag/drop work then? xD
by not sandboxing
That's just a big security risk then, that's what Flatpak is for
And native packaging is hard, devs shouldn't need to maintain 15 packages
This is why Flatpaks/Snaps exist in the first place lol
its not at all
this is why flatpak and snaps exist
sandboxing is an afterthought with already plenty of holes in it. discord for example can’t directly access my pictures directory which is super annoying, but if someone triggered an exploit they could just use the xorg socket to spy on all other apps, access files that they can, etc
it’s only actually secure when apps adopt apis meant just for flatpak which doesn’t really help in adoption
Ik, its annoying
Portals need to be used more
when I'm in a worst takes competition and my opponent is magma:
I dont see that as a negative. having a permission system is better than having nothing.
take android as an example, it adds new permissions with every update and while android is not the most secure system, it tries it best to protect users from shady apps
I consider flatpak to be the same. creating a new distribution system which takes efforts to also consider permissions is a lot better than most distro packages
when that permission system regularly
- annoys users
- breaks expected functionality
- gives users a false sense of security
i would say nothing is better
android permissions be like
Allow TotallyNotAppThatSellsYourData to access your contacts?
Deny
Allow TotallyNotAppThatSellsYourData to access your contacts?
Deny
Allow TotallyNotAppThatSellsYourData to access your contacts?
oh stfu already Deny
Allow TotallyNotAppThatSellsYourData to access your contacts?
aaaaaaaaaaaaa
take android as an example
you can't. android actually has a pretty well thought out, secure permission system that apps are literally required to use. flatpak doesn't, and instead just forces apps into that environment without any actually good way to expand those privilege. best you get is flatseal, which is not a great tool honestly
android is not the most secure system
it's probably only beaten by ios in security for consumer devices tbh
creating a new distribution system which takes efforts to also consider permissions is a lot better than most distro packages
and one day, sure. but i honestly don't think this day will ever come. it's linux, not everyone is going to switch to it and adopt, especially when some apps are just a horrible fit for flatpak and it's model (vscode)
the only 100% advantage flatpak has over (traditional) distro packages is giving developers the ability to use the dependencies they want, wherever
which is great don't get me wrong but like
i don't get the hype over a permission system that's really just annoying to deal with and filled with just promises
ios has been hit hard by this recently with some of the photos perms
i give discord access to whatever i select but any time i open it to send something it asks me for more
it's so fun
you can't. android actually has a pretty well thought out, secure permission system that apps are literally required to use.
no it took ages for it to become like that. iirc android 5 added the newer permission model and starting added a bunch of new stuff
before that it was just some basic permissions and apps were free to do whatever
while I agree, the current state is less than ideal (yes I also feel discord is annoying to use. drag and drop fucks up the entire app) but I see this as a positive change and eventually be better to use
i don't get the hype over a permission system that's really just annoying to deal with and filled with just promises
promises is all linux can do tbh given how fragmented the ecosystem is. hype is probably not worth it but I can see it being in a better place in the future ( as wayland ig? )
The user shouldn't need Warehouse and Flatseal if an app has a missing/incorrect permission
Why can't Flatpak do the same as iOS and Android?
Such a mess rn
Welp. Deal with it or get someone to verify the app then
lol
Mint's just trying to keep users safe
But Clem is also tired of Gnome's stupidness
She(?) wants all these GTK-based DEs to have core apps (notepad, scanner, pdf reader, etc.) via XApps so everything's consistent
its more that each DE shouldnt need to recreate their own suite of apps that behave normally and dont break when they are used outside their DEs
Exactly.
Is it just me or does the app not work on x11? It just gives me a white screen. Works fine on Wayland
Why use x11?
Lemme guess, Nvidia is having issues
yep
why does drag and drop not work with the flatpak of modrinth app for e.g. datapacks or resource packs in minecraft?
solved by giving acces to download folder
classic flatpak moment
wouldn't it make more sense to bring the download access ootb with it
it’d help but not really fix the core issue here
As a temp. fix it'd help
PRs welcome
fixed ;))
I don't think https://github.com/flathub/com.modrinth.ModrinthApp/pull/8 or https://github.com/flathub/com.modrinth.ModrinthApp/pull/6 are needed since https://github.com/flathub/com.modrinth.ModrinthApp/pull/3 was merged
Probably wrong but there's like 3 Gnome webkit/runtime PRs here
Lots of Gnome stuff lol
they are
Ah
I made a builder for tauri apps in nixpkgs, would it help package modrinth for nix?
It's not currently in nixpkgs but I am working on one.
a build-support?
OH you mean the tauri builder itself?
i thought you meant the app wasn’t it in nixpkgs
yeah I mean a builder like the any of ones in build-support.
no XD.
oh yeah that would be a big help. i’ve been thinking of making a builder as well since i maintain this and gitbutler (which also uses tauri)
I am making one called buildTauriApp which hopefully should support most of npm packages.
I am also going out of my way and improving tauri docs by adding documentation for nix & nixOS.
i’d love to work with you on that if you’re interested.
i’m not sure how much we can exactly share between packages given how many differences i’ve seen, but the main bundle build/install is probably what should be targeted
have you considered a hook instead of a dedicated builder? i think that could work really well as it would let us not worry too much about supporting (one of many) node package managers and make things a bit easier to compose (which in my experience you end up needing a lot with bigger tauri apps)
sure thing.
well it's really just a wrapper around rustPackage.
not really but I am thinking of adding an option to adding fetch deps.
but I wanted to make a builder to make it easier to build their tauri apps, as there is currently an issue about this: https://github.com/tauri-apps/tauri/issues/8588
which I am also working on.
This is what I have so far: https://github.com/Eveeifyeve/buildTauriApp I am using modrinth as a good example to try and get the builder working.
But I guess I could look at if you can send me the nix documentation on it.
But I wanted to make the builder as there is many tauri applications out there I want to support that use different node package managers.
there isn’t much documentation on actually creating them sadly. the fetchPnpmDeps pr is a good recent example i know of that can help you get the gist, though
along with all of the hooks in build-support listed here https://nixos.org/manual/nixpkgs/stable/#chap-hooks
and i’ll make sure to look at this later and dm :)
Well I could make hooks that setup the package manager for the builder?
Even tho you will still have to manage diffrent package manager hooks but I want to add the support for only npm, pnpm & bun
but if people want yarn & deno as a package manager support they can maintain it.
But I am supporting bun as I create minecraft client(AKA modpacks) launchers in tauri.
npm just because node installs it by deafult.
Well I got modrinth working with my buildTauriApp
on aarch64_darwin
Building for x86_64-darwin with Rosetta to see if it works.
it will take probally around 12 mins to build a package.
but in nixpkgs repo it's all cached
x84_64-darwin tested.
nobody cares but for arch (AUR) there now exists modrinth-app-appimage which uses the appimage release while modrinth-app-bin uses the deb release, i also now maintain the modrinth-app-git which was broken for some time
Not sure why everyone doesn’t just use the Flatpak, it works everywhere
Nix and Arch aren’t really that special, they’re just more advanced distros
it has problems with GPU switching
usually you just add a few variables and minecraft uses your GPU, this doesn't work with flatpak for whatever reason
well you wouldn't suggest that to an average user
Ik lol
ive never used any flatpaks
dont see the need to start using them either
Well, system packages are messy and you have to rely on the distro maintainers to keep things updated. Otherwise you’re stuck with packages that won't get updated for who knows how long. And yes, Flatpak may have its issues, but you're getting updates right from the devs (if they're verified then that's even better) and are much more up-to-date
@lavish relic If you wanna be stuck with insecure and outdated software, then that's fine. Be my guest ^

Wouldn't be surprised if you use Snaps too. They're just as messy as sys packages
no i hate those and one of the reasons i use debian instead of ubuntu
Good. Hell, Solus has been supprting Snaps for a while, but are gonna move to Flatpak in early 2025. They're tired of them too
At some point Snaps might be exclusive to only Ubuntu
Which ngl I'm not mad about, they do suck
And tbh sys packages might die too (or become very niche), Flatpak is supported by most distros
yeah permissions are so fun :/
time for another hacky bash workaround
you’re hardly ever getting updates directly from the developer and it’s actually common for things to be more outdated in flatpak, especially dependencies
there’s no reason to target people over this stuff
Just mentioning
hm? I use it fine for other flatpak apps tbf. prism and bottles work fine
Well outdated and insecure software is not bad unless it major 5 cve like you don't have to worry about updating the software often.

there's like banks and airports and stuff running extremely old windows versions and they're doing fine
and honestly who cares, they likely aren't actively being attacked by anyone anyway. you're just complaining for the sake of complaining, there was no reason for you to target them like that
and yeah most flatpaks typically have out of date deps as getchoo said :p
@cold dove was it you who shared a cross distro package search thingy on this server ?
a website to look up how different distros package a specific artifact. their names and repos
(sorry for the ping, but it was on prism server and I found it the repology website. regardless, thanks! )
@cold dove sorry for ping, from what i can see modrinth-app is supported package on nix?
yeah okey
0.8.2 will be on flathub once the bots pick up the build
a pr has been opened for nixpkgs as well
0.8.2 released?
ok seriously where are the git tags or releases or any notification about a release
where am i supposed to keep an eye out for a release
ok well 0.8.2 had a post in #dev-updates but not role ping (and 0.8.0 or 0.8.1 didnt even have that)
i guess ill build something around https://launcher-files.modrinth.com/updates.json then
0.8.1 had an everyone ping
though yeah 0.8.2 should've had a ping for the app notifications role
where
yeah im wondering that too since honestly i didn’t know 0.8.1 came out (or 0.8.0 if it existed). only reason i knew about 0.8.2 was because of the recent blogpost
0.8.0 was a series of pre-release versions in Labs
0.8.1 we soft launched a day before the blog post to get wider feedback before announcing it with the post
ah gotcha
forgot to mention: there may be issues with the 555 and 560 nvidia drivers and newer versions of webkit (i.e., those found outside of the appimage) that trigger a segfault on startup. i haven’t seen this on any mesa (amd/intel) systems, and using nvk fixes this as well
we might need to downgrade the version of webkitgtk used to fix this…which isn’t great since yk, it’s a browser
tauri 2.0 might fix this too. i’ll have to try out one of their demo projects and see
Hopefully 2.0 is out soon, cause it’s starting to block a lot of upgrades for internal stuff the app runs on which would fix a lot of issues
FYI: 0.8.3 has been soft-launched, so far no bug reports which is super promising
0.8.3 is now on tauri 2.0
awesome!
update: 0.8.3 and 0.8.4 have been pulled because the auto updater is broken lol
Would be nice if the updater didn't just re-use the MSI installer, Curse does this without one
Should mean less issues once all that gtk stuff is updated
(and client-side decorations are figured out, because Gnome is stupid)
why not?
Feels kinda odd imo
well it seems easier and cleaner to have one method of installation rather than writing an entirely separate one that does the same thing
Tell Curse that, lol
nah, msis are actually quite annoying and a source of all sorts of obscure errors on windows
I think having customised NSIS or InnoSetup installer that can do silent installs and updates would be a huge improvement, and also not using administrator to install the app too
what about microsoft’s newer packaging like msix
i’ve heard it’s good although i don’t really seem them used often or use them myself
the msix installer UI is at least fantastic
have no idea about the format itself
i've heard a lot of great things about it, i think at some point getchoo was working on an msix package for prism
last i looked there seemed to be almost no documentation on it, but at least something comes up when you google it now lol
Fixed my MSI installer issue by reinstalling Windows, lmfao. Something broke and I never noticed or could figure out
Only took 3 hours. Was not fun
it might be
some bugs in webkitgtk may still be there even after the abi change
but this should make it easier to package
since the abi tauri now targets isn't deprecated lol
it's...alright. it took me a bit of digging, but if you re-read the same sections enough in ms' docs you figure it out
i highly recommend it, though there are some catches
biggest is that code signing is a requirement (this is why i stopped development on it in prism. we aren't too confident in being able to have code signing forever given some recent events) and "debloated" windows installs won't be a fan
similar to msi here, a lot of these scripts/custom isos will just remove components of windows that are responsible for it
easy enough to fix though and obviously won't affect most average users
I stick to official MS stuff, and I've debloated 'enough' that doesn't cause stuff to break
Mainly, ChrisTitusTech's WinUtil and GeekUninstaller
neither of those are "official MS stuff" and can break your system
kinda my point :p
well maybe break is a bad term
but they give you the ability to remove and/or disable things that have no real reason to be disabled
By official MS stuff I mean not removing critical stuff from a Vanilla windows install, lol
Hell, MS shouldn't make critical Windows components so easily breakable in the first place
Look at an immutable linux distro and how hard that is to break, unless you're doing wonky things with it (and most of that stuff would be far more irreversible then a simple script that isn't doing wonky things)
what would happen if you do pacman -R pacman on linux
On a non-immutable Arch that would fuck things up, lol
they actually make it pretty hard
it's why people have to make entire scripts for it and constantly update them
Well, it seems stuff related to msi installers is very breakable, since for all I can guess the normal Windows 11 ISO I got had corrupted files in it, lol
Windows's SFC fixed them, now let's just hope it stays that way. Been smashing my head on a wall trying to fix this
@cold dove it's rolled out now!
the main problem atm is there seems to be some random segfaults on some systems. i can't repro tho
Thouyght 8.4 was the latest version, what changed in .5?
bunch of fixes
Ah
the Tauri v2 upgrade unsurprisingly had lots of problems
unfortunately the auto updater broke so anyone who updated to 0.8.3 will need to re-download from modrinth.app
Not surprising, think v2 was a big update
I was on 8.2, never got 8.3 for whatever reason
Unless the updater broke/wasn't detecting 8.3
probably didn't open it before it got pulled
Ah
Well, good thing I didn't lmao
And at least Getchoo doesn't have to rely on deprecated stuff now
nope, upstream bug lol
https://bugs.webkit.org/show_bug.cgi?id=278230
looks like canonical have been backporting webkitgtk updates to 22.04 as well, so this might end up hitting the appimage at some point
it's just one patch release behind right now
pretty sure that has some bugs on nvidia too though :/
https://packages.ubuntu.com/jammy-updates/libwebkit2gtk-4.1-0
||we should've just used electron||
for linux it is infinitely better
not to say there aren’t some hilarious electron regressions for linux too
but in general chromium is a better base than webkit on linux. idk why tauri devs decided to go with the latter
I'm not dealing with Electron unless its an app that uses it sanely and correctly, lmao
Discord is on my Deal With It list since no alternative exists that supports [Redacted]
(or however that's spelt)

Discord is using the same amount of RAM in Firefox, that's a problem with Discord, not Electron
Discord is just bad lmao
Electron is actually okay
Discord just uses it incorrectly, and I'm kinda surprised no one's ever bothered to fix it. But as per usual, money and spamming Nitro ads sometimes above anything else these days
Well tauri is much better for app size have a look at an app like spotify and discord and compare it to modrinth.
Since it's not shipping a chrome browser.
It uses native instead.
Switching will equal more app size, loss on security and loss on sidecar.
Is it really worth to switch, do you think?
Also to mention packaging electron apps on certain linux distros are pain in the ass.
Like nix.
Imo reason enouth for using tauri is not supporting chrome
.
.
This is really true, try packaging apps that are using electron on nix.
Tbh I don't see the point of Nix these days
Fedora's way of immutable distros, via RPM OSTree, is just easier to deal with
Especially with things like Bazzite and the guys at Universal Blue
I absolutely love Nix and during the time that I used both I have had a way better experience with Nix and NixOS than Silverblue
It seems like Silverblue is easier to start using but harder to maintain in the long run, NixOS is a pain to get into but is generally just a better experience once you get past all of the hard/annoying stuff
Nixpkgs is also amazing
Nix's terminal-centric-ness is what makes it very hard for new users
There is Snowflake, but I'm still just not a big Nix fan
Ostree‘s layering is getting slow by time. I do prefer transactional-update or Systemd‘s sysupdate here
Oh and I can also use Nix on MacOS
Just tried it earlier today and it's great
me when being able to install a package to the system without a reboot is an unstable feature
linux is awesome
That's immutable distros for you, lol
And either way, Distrobox is easier then layering packages, which could cause more issues if you use many of those
And tbf, this stuff is still pretty new
It's been a few years since its intro iirc
but it's not
nixos is immutable
it doesn't need to reboot to install, upgrade, or remove packages
Give RPM OSTree-based immutable distros like 10 years and they'll be much more easy to use
And Nix has a big drama issue, if you've seen Brodie's video that covered this
(Brodie Robertson)
can't wait
I'm not getting into that topic, go look that up yourself. I only mentioned since Ik its an issue
Same tbh
im well aware of it. i work on nixpkgs almost daily
really overblown
Agreed, but didn't know if you knew or not, but Ik about the packaging part you do
Then use guix.
Guix fixes most of these issues.
And is more maintained/less-experimental as to nix.
And been out longer.
The Guix package manager and the Guix System drew inspiration from and were based on the Nix package manager and NixOS respectively.
I think more people should tbh
I've heard of it a bit, but never really knew what it was
0.8.5 was pulled?
i think there's something going on with the referrer link
if i go to modrinth.com and click on "get the app" button, it'll give me 0.8.2 download links
going to modrinth.app gives me 0.8.5 links though
can't repro 
could use some help testing the 0.8.5 flatpak. im segfaulting from webkitgtk bugs with my nvidia card lol
anyone can install it with flatpak install --user https://dl.flathub.org/build-repo/127465/com.modrinth.ModrinthApp.flatpakref
do you happen to use wayland
Works for me here
yeah
myeah same problem on arch, appimage works fine but all the other packages segfault on wayland but work on x11
after updating todays kernel my appimage stopped working 😦 6.10.8-arch1-1
wayland or x11 ?
wayland, looks like it was a DE issue weird, works fine in gnome. Was experimenting while in cosmic, thats not a issue with you guys all good 🤔
well not an issue with the app but could be an issue with some arch dependencies / their versions
potentially, worked fine in hyperland / gnome. cosmic must be missing something or stuffing with x11 windows might bring it up on there git issues
does it really matter all that much? between the choice of having bigger app size and possibly some distribution pain and app not working at all I think the first choice is obviously better
webkit on linux is complete garbage. no clue why tauri people thought it was a good decision to use it
like the amount of reports we get of modrinth app simply breaking on linux in all sorts of ways is EXHAUSTING
sometimes it will run but performance will be horrible, making the app literally unusable. other times it will run but the window is blank. and in other cases it simply segfaults itself
‘try this env variable? doesn't work? awh, maybe try this one? doesn't work either. welp I guess no modrinth app for ya!’
is it possible to configure tauri to ship our own webkit version?
so it basically runs like electron on linux
I don't think it will help if webkit is breaking because it's incompatible with nvidia/amd drivers, wayland, or whatever
Can't the Linux version just switch to actual Electron or something?
Would probably help
the appimage does this already (using the version of webkitgtk in ubuntu 22.04 from ci). other tauri bundles like the deb can't really though, as webkitgtk is meant to be a shared dependency from the repository
the only other formats that allow you to do it would be nix and flatpak
but yeah it doesn't matter much when this is the case. the only solution would be to use an outdated version that hopefully has less regression cases (you already are sort of doing this), but that's not exactly a great idea for a browser
especially webkit with how often security vulns pop up
it would probably take a good amount of rewriting upstream, and given linux is linux, i can't blame them for not doing it lol
Eh true, but it would help with all this AMD/Nvidia/Wayland mess
Gnome also loves doing things in a needlessly different way that just makes things harder for everyone else for no reason
thats not how it works. if you use system apis for anything, all those would need to be replaced
I can see devs deciding to drop gnome support honestly. its getting close to tipping point
Exactly.
uhh is there a way to disable the built in auto updater
(trying to update the aur packages to 0.8.6)
you sure about that
but i wouldnt be able to do that for the binary packages anyway for the source packages yes thats a valid solution (its not actually even enabled by default i believe)
ok nvm that picture my mistake lets try again
also i love how the 0.8.5 git tag points to a 0.8.6 version
it's the latest
patch the tauri config
^
anyway i did a pr for that https://github.com/modrinth/code/pull/2388
that feels a bit silly
is it, i've seen it with other software
generally other software implements their own auto updaters where that’s the only way to disable it (node package managers for example)
i think this is more just a sign you should be compiling from source
(especially on arch where random binaries are fragile)
well yes the modrinth-app package builds from source and thats all fine but not everyone likes to wait for build times so modrinth-app-bin just downloads the release
i dont really understand what you mean by this
And the AUR in general is
This is why I suggested the Modrinth App Flatpak in the first place, and while Ik it isn't official, its the most stable way of using it on your Deck or whatever (once all this Nvidia/Wayland/AMD stability mess is figured out)
what i meant was that most software will only have an updater toggle at build or runtime. it’s rare to see both
and in this case it’s especially weird since it’s meant to apply to packaged binaries
there’s literally 0 use case for it upstream
flatpak isn’t much better for binaries, they just tend to deprecate things later. i actually had to compile a whole new version of webkit to get older app binaries to run
but hey, that’s linux for you. binaries suck unless you compile them for your distro
or runtime in the case of flatpak lol
Oof
technically true yea
but it does give users the option of not being forced to update
and makes 3rd party binary distribution easier
also is it it really that uncommon, using an env variable is uncommon for sure but many have some sort of switch in app settings
env vars ? Yes
most end users dont even know what env vars are. Even if they knew what they are, setting it up so that env is set up properly for that specific application or setup on global level is hard. No one wants to edit text files to play minecraft
it's (mostly) intended for 3rd party packagers like myself though
i dont see why anyone would oppose this harmless change that fixes problems
Well, its better then system or AUR packages imo
sorry for the lack of updates on the flatpak recently, a lot of things in my personal life have come up in the last few months
anywho, 0.8.7 has been pushed here. i would appreciate some testing as it seems webkitgtk is still hilariously broken on nvidia :/
you can install it locally with flatpak install --user https://dl.flathub.org/build-repo/132174/com.modrinth.ModrinthApp.flatpakref. modrinth:// links should also work now since we're on a version of the app that actually supports them
thanks all!
ehm... that didnt happen before
not sure if this is related to the flatpak but it didnt happen in the old version and building from source also doesn't cause that issue
running:
fedora 40
kernel 6.10.10-200
gpu NV160 - mesa 24.1.7 ( i think flatpak comes with its own mesa so...)
any idea how to debug that?
my guess is that it’s gnome’s version of webkitgtk
and maybe an issue with nvk, since i remember seeing some artifacting like this in other applications a long time ago
It not tauri decision as it goes from the native browser on your system and if it doesn't have a default then it goes to webgtk as it's defaulted across linux.
Yes they could have gone installing firefox which is another one that works on linux but it's not worth the fix.
But they said they were going to replace webkitgtk in v2 but didn't have the time for it.
#1277291712901021716 message
In tauri discord.
But yeah I heard in tauri v2 the issues are more bearable on linux as it uses a more up to date webkitgtk.
#1267210690431815863 message
webkitgtk is not defaulted on linux, from what I know qt uses cef (chromium embedded framework), and firefox is not good at being embeddable
they should've just made it a choice really, webkit is terrible on linux
Most linux that run gnome under the hood.
(x) doubt
eg. ubuntu
Well they are looking into replacing webkit as above I linked to some comments from fabianlars.
But they don't have the time to do it.
As they would have to recode the wry backend of tauri.
These are the most huge parts they would have to recode.
But yeah it's about whether they have the time @livid tiger.
Also don't forget to mention if it's a webgtk issue then try to report it to https://github.com/tauri-apps/wry/issues
it's actually their own thing called qtwebengine that's sorta like cef but not (i don't why they don't just use cef, but hey)
they can't do anything about webkitgtk issues. the ones described above come directly from upstream, and should only be reported there
True but if it's the actual runtime where they messed something up with webkitgtk in their code it's always helpful.
And the only bad thing is they will tell you to report it somewhere else and close your issue.
I thought it was just a wrapper around cef to create qt browser component
Especially if Gnome is the one controlling it. They made it for their stuff, and their stuff only
Most of the time its Gnome's fault ^
lmao
lol
welp apparently the appimage is just broken
also still love how the executable name changes like every other release
it changed only one time in.. 0.8.2?
modrinth-app -> theseus_gui -> modrinth-app -> 'Modrinth App'
0.8.8 with the space is especially annoying as the icons are also with a space and thats invalid so gtk-update-icon-cache throws an error
Things on our radar for the future we feel we should mention at least:
Providing or Bundling Chromium Embedded Framework (CEF) for Linux as an alternative to WebKit2GTK
yoo
this will be huge
can we get my pr merged before next release? idk why it wasnt merged for 0.8.8. would really appreciate it
now we'll only need to worry about hw accel breaking 
welcome to linux, all gpus will have random bugs with software
but yeah cef will be massive. hopefully we won’t need to wait til 3.0 for it
Really wish Linus would ask Nvidia face-to-face about open-source drivers tbh
Ik there was the one that the guy got hired for but since then there's not really anything going on
I'm aware that Nova is gonna(?) be a thing but, still
les goooo
nova is the RH thing yeah ?
nova + nouveau/nvk would hopefully be awesome
Yes
Not sure what the Huh reaction was for lol
Tons of 0.8.8 stuff
wow flathub’s automation is…something
The last PR is mine, not Flathub bot’s. The other ones were opened because the AppImage was updated on Modrinth without changing the version.
oh yeah all the checksums are different for the bots
that’s a little weird
guessing it was #dev-updates message
Should probably merge those
3 of them can't
the last i will review when i have the time
the modrinth flatpak is not the only thing i work on
0.8.8 is on flathub
Holy
Almost 21k installs
Not sure where all of those came from lol. Guessing Steam Deck
31k downloads total and ~200/day on average
https://klausenbusk.github.io/flathub-stats/#ref=com.modrinth.ModrinthApp
Pretty amazing still ngl
my modrinth app still doesn't launch, any ideas (version 8.8, (flatpak))
have you tried reinstalling?
the debug information would probably be able to help more, though
@shadow kite ^
no logs in the flatpak run terminal & in the debug info, reinstalled app & ~/.var/com.modrinth.ModrinthApp deleted so basically full reset
on all my devices (amd, intel)
flatpak update --commit=2c272ac8480cc6dc635451bae84636949472326662b8d3dd33a532b20d0df719 com.modrinth.ModrinthApp this is a temporary workaround for me
“Invalid link across device boundaries?”
Yes, no idea what this should mean
thanks
do you have multiple disks mounted at different places it could be using?
because that error usually means a symlink is trying to be made across two different disks
if it’s not that i’m guessing the launcher is trying to symlink something and the way flatpak bind mounts certain directories flags this
nope just one disk
https://github.com/flathub/com.modrinth.ModrinthApp/pull/28 this pull request fixes the bug for me
The issue is most likely with the auto updater trying to install the newest version in the read-only Flatpak file system. Since my PR uses the latest version, there’s nothing to update, but it will most likely break when a new Modrinth app version releases, as it has been for each release. Unfortunately there’s currently no way to disable the auto updater.
and thats exactly why my PR should be merged
Whose PR is which I am confused. Both are talking about "my pr"
merged. thanks for linking it
Most distros have a GUI app manager, Flatpak already relies on Gnome Software/Discover and whatnot and handles updates for you
Modrinth should rely on these app stores and not have its own update system, thus not needing an env. var. at all
@cold dove Sorry for the ping, just wondering why this needs to exist ^
the app doesn't rely on appstore, though. the team distributes it as an appimage, which isn't going to be touched by any package manager
Flatpak does though
yeah but the flatpak uses the appimage since it can’t build from source
Ik, that's an issue if Flatpak already has an update system
well aware
flatpak failing again because of 0.9.0 release, please merge the pr from the bot i'm gonna test in a few seconds
why isn't this merged yet 😭
yeah it works great
the new design takes some getting used to

thanks btw for this great request
it's amazing how a proper couple line improvement can take months and months to merge when there are such great merged prs as "feat: hawk tuah" for the servers stuff
Especially when 28,000 installations can cause problems unless this get merged.
only popularity, but that's not useful at all
Guessing Getchoo's busy, hope he'll get 0.9.0 out on the Flatpak side of things soon
Let’s hope that this is getting merged for 0.9.1
bump has been merged. waiting on flathub to publish it
See https://github.com/modrinth/code/pull/2388 if you haven't
Ah
nixpkgs bump is up
https://github.com/NixOS/nixpkgs/pull/367803
and not sure if i mentioned it here, but binaries built from nixpkgs are in the nix-community cachix bucket now
do you have an idea why the app is so laggy now on linux
no problems on flatpak for me, i guess it's because these other packages use not the greatest webkitgtk versions. The webkitgtk in the flatpak is isolated so no problems there
I think we should put much more emphasis on the flatpak because in my experience it works best of all the Modrinth packages. (apart from that https://github.com/modrinth/code/pull/2388)
Tested on x11 and wayland
like most linux things, i’d guess it falls on webkit. did this start around the tauri 2.0 bump? or is it more recent?
Tauri 2.0
We upgraded tauri
that’s what i figured. you’re probably stumbling across issues with the newer webkitgtk version from ubuntu 22.04
newer releases have had regressions left and right. tauri v1 and ubuntu 20.04 saved you for a bit since it used a way older version (this is also why distro packages had issues the appimage didn’t for a while, like with nvidia)
not much to really do about it except wait for stuff to be fixed
only problem is that ubuntu 22.04 may hold back those fixes to maintain abi compat for that release
if/when that happens, you should be able to bump your ci to 24.04 though and that’ll give the appimage any new version
keep calm and blame tauri webkitgtk ¯_(ツ)_/¯
Gnome being Gnome
i must say it got way better with gnome 47. epiphany is kinda usable now, beside from extensions
True, but they do things in a very Apple-style way
these are technical bugs
the core gnome desktop team is also very separate from webkitgtk
a lot of the issues also stem from webkit itself, which is basically maintained by apple
they can’t easily use firefox, it’s horrible for embedding
and generally apple is considered a more open maintainer than google with chromium (kinda evident by the community having basically full control of an in-tree port with webkitgtk)
- chromium monopoly blah blah
tauri will hopefully adopt cef soon though. i know they were talking about it
it still has regressions from time to time, but they’re quickly fixed and not usually as severe as the entire app segfaulting (or ever that severe, really)
Guessing cef will fix some of this stuff?
tested works
0.9.1
deb 0.9.0-2
rpm 0.9.0-1-1
???
huh xD
the 0.9.0-2 one is indeed 0.9.0 and therefore doesnt launch because it tries to auto-update
that means i cant update the aur package which also doesnt launch because its stuck on auto-update
<3
i need https://github.com/modrinth/code/pull/2388 so bad
dont we all
there is also no git tag for 0.9.1 so i have to rely on commit hash again :)
would be useful for scoop too
I still need to fix my bucket, it got stuck on 0.8.5 because installer is bad again now, and I need to extract from it instead ;-;
yoo we got a git tag
oops will fix
can i also ping about getting 2388 merged
Still waiting on the 2FA third-party sign in issue I wasn't aware of at first
#1316633256346783754 message
https://github.com/flathub/com.modrinth.ModrinthApp/pull/34 this works too of course, just tested it
I have already prepared a small pr so that it can go very quickly when 2388 is merged ;). Of course do not merge before: https://github.com/flathub/com.modrinth.ModrinthApp/pull/35
merged, thanks for testing
also if you have a nvidia card, issues should be fixed on 565 by setting both WEBKIT_DISABLE_DMABUF_RENDERER=1 and WEBKIT_DISABLE_COMPOSITING_MODE=1
or at least they were for me the other day with gitbutler (another tauri app)
the former seesm to really hurt performance but hey, least the app can actually start
#modrinth message
This should finally solve that though
This is great and bad at the same time, I don't want to support Google's monopoly but unfortunately it just works better, it would be great to have the ability to choose Firefox's Gecko.
Agreed.
Hi, the auto-update just broke my AUR version, so I removed it and installed the flatpak version in the hopes it would fix some of my issues, but now I only get this without it actually opening:
# flatpak run com.modrinth.ModrinthApp
Gtk-Message: 16:07:56.939: Failed to load module "xapp-gtk3-module"
Gtk-Message: 16:07:56.977: Failed to load module "canberra-gtk-module"
Gtk-Message: 16:07:56.985: Failed to load module "canberra-gtk-module"
just switched to the modrinth-app-appimage and that works, but still with the 2spf from libwebkit lol
basically unusable
all the aur packages are up-to-date tho, did you even try just updating xd
yeah i tried just updating and it just said it was up-to-date
well now it works, just that I can't really browse/add mods because it would take ages for that to render
lets just pray we get chromium soon 
Or Geko from Firefox
ended up trying this on my end. doesn't work with the app, but does with every other tauri app like gitbutler and one of their demos
considering they're both using the same webkitgtk and everything, i think there might be something about the app specifically causing this
huh. so there are cases where those dont even work....
well thats annoying, i thought it was like, dont use nvidia? yey things work. use nvidia? use this env var and suffer from lag.
that seems to be the case with other tauri apps
at least in limited use
it’s totally possible gitbutler, etc could segfault at some point as well with extended use, and modrinth is just taking that code path (or a similarly buggy one) immediately at launch instead
since honestly i’d be surprised if the app is doing something directly to cause this, just using something that’s buggy
2388 merged 
finallyyy
Yay.
Not a bad joke, despite its issues, I agree
Well, most of the issues are Webkit hating Nvidia
So cef or something would be better
Ik Chromium isn't that great, but if it fixes all these crashing issues, then its better then nothing
we doing this again ?
0.9.3
SIKE
<0.9.2
(linux deb, darwin x86_64 and aarch64)
oops sorry will fix
but
thanks so much we finally have MODRINTH_EXTERNAL_UPDATE_PROVIDER
also missing a git tag
if anyone would like to test the flatpak
flatpak install --user https://dl.flathub.org/build-repo/162204/com.modrinth.ModrinthApp.flatpakref
the version number is correct in the url now but the file is 404
- thanks for the git tag <3
edit: file is now found <3
Is this with an updated Tauri build?
Or is cef still not used
still webkit
Ah, oof
this is just a repacked appimage
no version of tauri has cef anyways yet afaik
Ah, oops
yeah, still doesn't launch on nvidia (pop_os)
with or without WEBKIT_DISABLE_DMABUF_RENDERER=1 ?
@grand shale ^
nvidia + webkit2gtk is a bad combo and thats why, add that env variable and it will launch but lag like hell
oops ig i should have pinged @grand shale
it currently launches but lags a ton
flatpak doesnt launch at all
its tecnhically usable but i just ended up using prism launcher
fair, this will be the reality until we get cef or webkit gets magically fixed
yeah i figured
in any case, the functionality of prism launcher to select a wide variety of mods and disable/enable them all at once is nice so ill probably stick w it till modrinth has that
purely because troubleshooting it much easier, since i can binary test my mods to see which is causing the issue much faster
xd more and more people switching away from windows these days
-# year of the linux desktop ? /s
we just need to hope that they choose the right distribution and one that isn't buggy or base their choice and opinion from websites that are sensational and always advertise as this as best gaming the OS Bs
🌎 Get exclusive NordVPN deal here ➵ https://NordVPN.com/pewdiepie It’s risk free with Nord’s 30 day money-back guarantee!✌
Get an exclusive 15% discount on Saily data plans! Use code pewdiepie at checkout. Download Saily app or go to https://saily.com/pewdiepie ⛵
Official ƿ૯ωძɿ૯ƿɿ૯ TikTok: https://www.tiktok.com/@pewdiepie
∞ Official ƿ૯ωძɿ૯ƿ...
|| Yes, your eyes aren't deceiving you. He actually used Mint.||
yeah saw, amazing
And the Modrinth App Flatpak's Tauri issues get fixed. And Nvidia.
will probably update the info tomorrow
pinging myself so i remember @honest cedar
@honest cedar Reminder ping ^
Eh, there's only really 3 base distros: Arch, Debian, and Fedora
Ik they aren't exactly noob friendly, but stuff like Ubuntu, Mint, PopOS, and Bazzite are all very good
Plus SteamOS and CachyOS
updated
Opensuse 🤓
I guess Gentoo users will have to fall back to building from source, which is fitting giving how the distro works anyway

Or compile Flatpak and install the Flathub repo
Then you're done lol
git tag for 0.9.4 pls 🙏
appimage broken? gives "Could not create default EGL display: EGL_BAD_PARAMETER. Aborting..."
ok nvm that is some other issue
I doubt this will get an update till Tauri provides other web options so Nvidia people can actually use the app
Cause rn its utterly unusable on an Nvidia card without a lot of tweaking that only half-fixes the issue
And AMD has some issues too iirc
Done
The Nvidia thing can actually be fixed with a single env var, Alex (our new app developer) figured this out
Ah
Just set WEBKIT_DISABLE_DMABUF_RENDERER=0 (you may need to force-override it in /usr/bin/modrinth-app)
This should be enabled ootb on the Linux versions then
Aha, but on certain other GPUs it only works if you set WEBKIT_DISABLE_DMABUF_RENDERER=1
So we can't set it for everyone
Could detect the GPU then?
Might be the only way to fix this issue for now(?), but mk
He has a masters degree he'll figure it out
I only know that I know nothing
that's a genuine sign of education though!
usually the problem is when people are unaware that they know nothing
could that get fixed
also is there someone i could/should ping about issues with things like missing git tags or files in updates being wrong
for this particular issue I think the only one who can fix is @rich shadow? is it just that the file is named wrong
git tags is something I can fix
didnt check if it was just filename or if its actually the wrong version
its actually the wrong version, not just named wrong (at least deb, idk about rpm)
@gleaming trench can also fix
what is the problem
is it the wrong file uploaded?
if so where do I get the right one, from GitHub actions?
should be fixed now I think?
👍 the 0.9.5 deb file is now available but just not listed in the updates.json file
the url is incorrect in the metadata file
it is now, or are you saying that's what the problem was? bc I did update them
probably meant that it was
yeah it was
sorry
There was at least one GH issue for this I think
Maybe this one too?
https://github.com/modrinth/code/issues/3609
@rich shadow
I believe 3609 is caused by the same root cause yes
Me too, Prospector already posted a comment in there so people can confirm
@rich shadow Many people complaining about links being broken in https://github.com/modrinth/code/issues/497#issuecomment-3064972688, think AppImage or others as well
Sorry for the ping
@desert apex another rpm link thing
it's a cache thing, not sure why it hasn't updated by now, kinda ridiculous
We have been setting a cache lifetime of a whole month before for the affected metadata file, this kind of random stale cache things can happen until then
AppImages may be a bit of a legitimate thing to investigate though because my browser at least tries to display them as text, instead of downloading them
May be a MIME type issue
On a VM to test but Modrinth App on Debian 13 refuses to work (Flatpak)
sigh.. @honest cedar Since I’ve been waiting for like a month with no response still, guess I’m pinging you
did 3960... just regress and remove the external update provider env variable.....
hmm, will ask the backend dev who worked on it
yeah, seems to have been an oversight when reimplementing the updater
will get that fixed before releasing 0.10.8
great! thanks :3
thanks for catching that
should be out now with that added back in
Sooo is the Debian issue fixed..?
you just got plain ignored xd
Pretty much ugh lol.
it’s a vm, so could be a driver issue with that
really not more info i’d imagine anyone can give based on this screenshot alone. it’d be more useful to run it from the cli and send what it spits out
i’ll make sure to look at it soon
i honestly kinda really slowed my foss stuff in the last month or two, and this is just another victim of it lol
someone came along and did all the hard work for me (and even added all the release notes!)
im also setting up codeowners now though so i'll actually be pinged about these changes
fyi, now that we have fixed the updater issues on Windows and Mac, we've been pushing way more releases, pretty much as soon as there are changes to the app merged to main.
does it currently take a whole pull request every time we do a release for the flatpak to be updated?
yes
at minimum the url pointing to the binary and the hash for that binary need to be updated
Glad to see Linux finally being taken seriously now.
lol this is how i find out i no longer need to maintain the aur packages
aaandd its finalized
Holy shit! Big work, thank you so much for helping with packaging and working towards getting this accepted in main repos



