Eden - dev-chat

I've been wondering about this for quite a while.

9:01 AM

How possible is it that shaders can be recompiled ahead of time if you can provide the necessary information beforehand (such as where their binaries are located)?

in theory yes, practically it'd be quite difficult to do in a manner that doesn't suck since every game/engine handles it a bit differently

hi everyone, I or rather GPT Sol 6.1 made a game work on my Samsung S11 Ultra with MediaTek Chipset and Mali drivers. Am I late to the party or may I share what we discovered? I want to point out that the solution to my specific issue was a one liner someone else wrote. It just happened to work for me

Preliminary Android/Mali finding: Super Smash Bros. Ultimate running in Eden on a Galaxy Tab S11 Ultra. Device: SM-X936B, Dimensity 9400+, Immortalis-G925, Android 16. Sources: Eden de7e6e01d4 (modified, separate debug APK); PanVK-kbase RC3, Mesa 5a07217f03. AI assistance was used for investigation, local code changes and build orchestration. Gameplay was manually verified by me. Controlled comparison using the same toolchain/settings and a fresh shader cache:

Rebuilt, unchanged RC3: CSF fatal status 0x7d0002c1, sideband 0xfffae000, followed by VK_ERROR_DEVICE_LOST before the title screen.
Same build with only the GPU helper-address temporary in get_fb_descs() changed from uint32_t to uint64_t: reached character selection, loaded a stage, completed the countdown, and produced a controllable match. Credit: this address-width fix was already documented in emir34-34's DroidDeck Mali fork. This is additional validation on G925/Eden, not a claim of original discovery. Eden also needed a targeted direct-ICD loading path for PanVK. The existing platform-loader/HMI route failed with VK_ERROR_INVALID_EXTERNAL_HANDLE. Shader generation and feature checks were unchanged. The direct path uses the port's CPU-copy Android presentation. One pipeline worker was held constant in both runs; it is not established as necessary. Observed approximately 60 FPS in menus and 38.7 FPS in a match while 10 shaders were compiling. Only a short gameplay test has been confirmed; sustained performance, other stages and long-term stability remain untested. The driver switch bypasses the original proprietary ARM compiler stall; the one-line fix addresses the separate PanVK GPU fault. This is preliminary local research, not an official or universal compatibility claim.

I was notified that Eden isn't handling display servers correctly, and this was blocked by a developer here (@crueter?) Is it also true that Eden recommends AGAINST using Wayland due to an already-fixed bug? This is a subpar user experience, spreads outdated information, and discourages actually fixing bugs that do crop up. And to address the display server handling problem, Eden should be going through this flow. It shouldn't result in a crash if Eden is told to use X11 or Wayland and can't find it for whatever reason.

- check xdg session type - if wayland, try wayland socket - if that doesn't work, check for x11, try that - if that doesn't work, bail out - if x11, try x11 socket - if doesn't work, check for wayland, try that - if that doesn't work, bail out

8:39 AM

This is a robustness issue, not some "feature". I would see this as a bug that Eden doesn't handle this correctly.

Leandro.

How possible is it that shaders can be recompiled ahead of time if you can provide the necessary information beforehand (such as where their binaries are located)?

correct me if i am wrong, but iirc this is how shaders are usually handled on many modern PC games, (large AAA ones at least), when they first launch they compile all the shaders which takes forever (such as in black myth wukong) and store them in the shader cache. Whereas on consoles, since all nintendo switches have the exact same hardware (same gpu) they just precompile the shaders for that console's hardware which eliminates shader stutter.

Is it also true that Eden recommends AGAINST using Wayland due to an already-fixed bug

We recommend against (and, on the AppImage, usually force disable) Wayland due to random crashes and performance issues that don't happen at all on X11

And to address the display server handling problem, Eden should be going through this flow. It shouldn't result in a crash if Eden is told to use X11 or Wayland and can't find it for whatever reason.

If X11 isn't available (bad idea regardless) you can set I_WANT_A_BROKEN_WAYLAND_UI=1

discourages actually fixing bugs that do crop up.

None of us use Wayland in the first place, we can't guarantee support especially when it is this broken; and likely will stay that way until someone volunteers to fix it, instead of being the 1200th person to complain. (edited)

If X11 isn't available (bad idea regardless) you can set I_WANT_A_BROKEN_WAYLAND_UI=1

This doesn't address my concern. Eden should not be silently aborting without letting the user know.

12:49 PM

If that means it goes to Wayland and shows a warning dialog, that's fine. But it shouldn't just fail to start.

12:51 PM

I would also be interested in hearing where Wayland issues have popped up and hearing what you have done to solve those rather than chronically put off debugging and fixing them. X11 is not your savior, and people are going to want features that X11 does not provide at some point (HDR if it involves Switch 2 titles at any point, fractional scaling, especially with multiple displays)

12:51 PM

Are they AMD GPU driver issues? Compositor issues? Protocol violations from something Eden is doing? NVIDIA problems?

They're issues across every tested combination of GPU vendor, compositor/DE, and have persisted across multiple Qt versions

People will especially start ending up in situations where you're forced to deal with Wayland, as developers, as they hit all of the problems I listed above.

AMD GPU driver issues? Plenty, just ask anyone who has tried rabbits/xc/or any of those games. Usually with wayland the problem becomes excerberated (i.e devise lost).

The most common case is unexplained crashes, UI lockups, device loss errors, and segfaults that don't occur at all on X11. Some compositors also had tanked performance due to lacking fifo-v1, though it's been long enough that's not really much of an issue anymore (I think there are some that still don't have that proto but idr)

lizbeth86

AMD GPU driver issues? Plenty, just ask anyone who has tried rabbits/xc/or any of those games. Usually with wayland the problem becomes excerberated (i.e devise lost).

In order for something to be exacerbated, it has to exist in the first place.

12:54 PM

Which means there's something to fix.

12:55 PM

all of those issues you list out sound more like Eden issues, and especially fifo-v1 problems are something that clients already work around (GTK, for example, does so by setting an EGL swap interval of 0 and pacing itself using frame callbacks)

12:56 PM

though that's less of an issue if Mesa would actually use fifo-v1 for EGL, which I need to bother some Collabora folks about.

1:02 PM

Anyways, my point being, it sounds like these are problems that actually need to be solved and forever putting them off isn't really going to help anybody, especially with KDE and GNOME going Wayland-only soon. And more and more users looking to drop XWayland, or having issues with X11 apps that are solved by switching them to Wayland (see: fractional scaling, HDR)

Oro

Anyways, my point being, it sounds like these are problems that actually need to be solved and forever putting them off isn't really going to help anybody, especially with KDE and GNOME going Wayland-only soon. And more and more users looking to drop XWayland, or having issues with X11 apps that are solved by switching them to Wayland (see: fractional scaling, HDR)

Yes some issues are excerberated, some are new through and don't happen on X11, which as said nobody uses Wayland on the team. I think you really overestimate how much the team is able to handle at once. We already struggle with AMD/NVIDIA bugs on X11, Android[qcom/mali] and Windows (at the same time). Done so for the fact we only have like 1 active GPU dev at a time. I don't really think it would be a good idea at all to add Wayland to the mix. So for the forseeable future Wayland graphical issues may just be unfixed just due to the lack of manpower to handle them, simple-as. (edited)

I see.

1:10 PM

I will point out again that this does not address the problem of handling falling back to another display server properly.

Good news, the hook that handles this is open source https://github.com/pkgforge-dev/Anylinux-AppImages/blob/main/useful-tools/quick-sharun.sh#L3132

Anylinux-AppImages/useful-tools/quick-sharun.sh at main · pkgforge...

AppImages That Truly Work On Any Linux System (Even musl-libc and non-FHS systems) - pkgforge-dev/Anylinux-AppImages

This is an Eden problem, not an AppImage issue.

1:12 PM

Eden itself does not properly handle the case of not being able to connect to the display server as configured (or for the session type), and falling back to the alternative.

That's handled by Qt

Eden is overriding QT here.

1:26 PM

    if (GraphicsBackend::GetForceX11() && qEnvironmentVariableIsEmpty("QT_QPA_PLATFORM"))
        qputenv("QT_QPA_PLATFORM", "xcb");

1:26 PM

src/yuzu/main.cpp, line 159 in main

If you turn on an option that isn't on by default, yes

1:38 PM

That also won't trigger on the AppImage unless you very much know what you're doing, and explicitly override the hook

this is an issue that a regular user can hit and have absolutely no clue on how to debug it

1:42 PM

and wouldn't even be able to fix it if they can't figure out why their X11 server isn't working (or isn't available), or doesn't know that they need to change Eden's configuration files

1:46 PM

you would still hit this even if you used the AppImage

You wouldn't hit this on the AppImage because it sets QT_QPA_PLATFORM to xcb anyways

not if your X11 server is fucked

1:50 PM

or otherwise nonexistent, or inaccessible

1:50 PM

the problem here is exactly that QT_QPA_PLATFORM is being set

1:51 PM

and Eden doesn't try to run again without it set if that fails

Right, and as you've been told several times, everything involved is open source

If your X11 Server is fucked I think one has bigger issues

One solution would be to set QT_QPA_PLATFORM to xcb;wayland which would try X11 first and only use Wayland as a fallback

that would work

1:59 PM

aye, can confirm, xcb;wayland does the job

1:59 PM

with an artificial nosocket=x11 test

2:00 PM

bonus points for Eden being able to drop warning dialogs and the like, if you do

    if (qEnvironmentVariableIsEmpty("QT_QPA_PLATFORM"))
        qputenv("QT_QPA_PLATFORM", "xcb;wayland");

2:02 PM

then it should only ever get forced to anything else if the user themselves requests anything else

lizbeth86

AMD GPU driver issues? Plenty, just ask anyone who has tried rabbits/xc/or any of those games. Usually with wayland the problem becomes excerberated (i.e devise lost).

add this recent issue to the mix https://github.com/ValveSoftware/gamescope/issues/2261

Samueru

add this recent issue to the mix https://github.com/ValveSoftware/gamescope/issues/2261

Oh im 100% familiar with that

lizbeth86

Oh im 100% familiar with that

See also https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16449 And the original responses I got from mesa were not buying what I was saying lol

lmao

wayland is such a meme that there is now legacy-wayland in mesa

11:19 PM

and they broke a ton of devices without giving a fuck

why do people even use wayland

Samueru

wayland is such a meme that there is now legacy-wayland in mesa

Similar to how support for X11's DRI2 protocol was deprecated in 24.2, begin deprecating EGL_WL_bind_wayland_display (including eglBindWaylandDisplayWL et al) by moving it behind a legacy-wayland build option. This extension was originally created in a pre-dmabuf world, where we didn't have a universally-accepted way of exchanging buffers between client and compositor, or even really the ability to describe formats and modifiers universally. Since then, the world has settled on dmabuf with DRM FourCC and modifiers. We've had the zwp_linux_dmabuf_v1 protocol for 10 years now: both clients and compositors implement this protocol to handle buffer sharing. Compositors either use EGL_EXT_image_dma_buf_import or the Vulkan dmabuf extensions to import these into GPU world.

11:52 PM

please do dig a bit deeper into what you're linking before saying things like this.

Oro

please do dig a bit deeper into what you're linking before saying things like this.

??? so what should the person that is having this issue do? Note this breaks all of EGL, not just what the PR mentioned

Samueru

See also https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16449 And the original responses I got from mesa were not buying what I was saying lol

@Oro read all of this

Samueru

??? so what should the person that is having this issue do? Note this breaks all of EGL, not just what the PR mentioned

applications need to migrate off off of those deprecated EGL extensions, there are better options available.

nooooo

A compile-time flag is provided to enable those exts if you really need them.

Oro

applications need to migrate off off of those deprecated EGL extensions, there are better options available.

read the actual issue report, the problem is not the applications

11:54 PM

this also breaks devices with split DRM topology like some rasberry PIs

11:54 PM

again read the issue

11:54 PM

it breaks all of Qt and eglgears_wayland

Samueru

this also breaks devices with split DRM topology like some rasberry PIs

that is not clear from the issue report (edited)

11:55 PM

I can't see at a glance whether your AI-generated likely-cause was confirmed.

Oro

I can't see at a glance whether your AI-generated likely-cause was confirmed.

11:57 PM

yeah wayland will get nowhere at this pace lol

I'm having troubles parsing that linked report.

11:58 PM

keeping in mind that nobody else has even came to look at this issue, and thus not even mentioned if there's a better solution to fixing this issue than relying on those old EGL Wayland extensions.

11:58 PM

which I'm willing to bet that there is.

11:59 PM

treating Wayland like some scapegoat is definitely a way to make sure it gets nowhere.

Oro

keeping in mind that nobody else has even came to look at this issue, and thus not even mentioned if there's a better solution to fixing this issue than relying on those old EGL Wayland extensions.

people don't report this issue because they don't realize they are using llvmpipe this problem is much worse for us because our builds of mesa only have softpipe and it becomes obvious with any app lol

this does not change my statement

12:01 AM

what help does it provide if you treat Wayland as something that doesn't deserve any bug testing or bug fixing, and talk down about it at every opportunity?

12:01 AM

How is this conducive to improving the Linux desktop?

12:02 AM

All you're really doing is making sure users are stuck with an increasingly-rotting display server that barely gets attention.

12:02 AM

and that's missing several features that users do want.

Oro

what help does it provide if you treat Wayland as something that doesn't deserve any bug testing or bug fixing, and talk down about it at every opportunity?

are you fucking kidding? I even opened this issue because I noticed a pattern https://github.com/pkgforge-dev/Anylinux-AppImages/issues/927 I don't have aarch64 devices, I had to ask someone else to do all the tests for me yet apparently my report wasn't clear enough

imagine a world... where only X11, peaceful rosemaries everywhere and SDL with proper swrast so no need to bundle mesa oh the peace.

Samueru

are you fucking kidding? I even opened this issue because I noticed a pattern https://github.com/pkgforge-dev/Anylinux-AppImages/issues/927 I don't have aarch64 devices, I had to ask someone else to do all the tests for me yet apparently my report wasn't clear enough

why are you taking my question as an insult?

12:14 AM

I'm pointing out behaviors and breakdowns in communication that are not productive.

lizbeth86

imagine a world... where only X11, peaceful rosemaries everywhere and SDL with proper swrast so no need to bundle mesa oh the peace.

SDL fixed this issue btw

12:14 AM

xd

Samueru

SDL fixed this issue btw

Yeah and debian will take 6 billion years to ship it

Oro

why are you taking my question as an insult?

said the person that just recently told me to dig a bit deeper after not reading the bug report lol

12:16 AM

but you are right

12:17 AM

maybe I shouldn't waste energy reporting issues related to wayland

your bug report has information spread out across two, three different issues, and it wasn't immediately clear to me looking at your Mesa PR why the issue occurred, despite you acting as if it's so clear

12:18 AM

if that's how you want to take things, sure, go for it. I don't expect you to be any more productive here.

Oro

your bug report has information spread out across two, three different issues, and it wasn't immediately clear to me looking at your Mesa PR why the issue occurred, despite you acting as if it's so clear

so how should I hav worded the bug report?

if you knew what the issue was (some driver fuckery), you could've laid it out in the issue (oh, vc4/panfrost/whatever are breaking and this compile flag seems to work around it, here's some information we got from our testing and we think it might be this specific issue), instead of spreading that issue out in other issue trackers and blaming the issue on the removal/disabling of deprecated EGL extensions that had alternatives as clearly laid out in the pull request you blamed the problem on.

Oro

if you knew what the issue was (some driver fuckery), you could've laid it out in the issue (oh, vc4/panfrost/whatever are breaking and this compile flag seems to work around it, here's some information we got from our testing and we think it might be this specific issue), instead of spreading that issue out in other issue trackers and blaming the issue on the removal/disabling of deprecated EGL extensions that had alternatives as clearly laid out in the pull request you blamed the problem on.

alright, will update the bug report with more details instead of pointing at the original debug session

12:29 AM

I mean, anything helps. Whether it gets fixed depends on maintainer free time and responsibilities.

12:30 AM

I did mention the issue to some other folks so maybe they'll put it higher up on their timetable.

12:30 AM

aarch64 hardware in general is just plenty of fun.

I also don't know what driver this affects exactly, since the behavior depends on the drm topology this is why I pointed at the original bug report

I might be able to test on an rpi4, 3 series

12:31 AM

unsure

12:32 AM

I'd have to find my hardware and a working display to set it up on, and I don't think I have anything for that

12:34 AM

I did see this in the issue report, so may be worth copy-pasting over directly:

Add -D legacy-wayland=bind-wayland-display to the mesa debloat scripts and rebuild the packages/AppImages. Restores the old fallback; kmsro is already built for the aarch64 mesa (zink pulls it in on 26.1+, the ARM gallium drivers before that). Upstream: fall back to device->nodes[DRM_NODE_PRIMARY] in the feedback handler when there is no render node. No MR for this yet (!38079 only touches loader_get_user_preferred_fd). Workaround: WLR_RENDER_DRM_DEVICE=/dev/dri/renderD128 so the compositor's main_device has a render node.

(edited)

12:36 AM

hm, I would be interested, just out of pure curiosity, if Tegra hits this

Oro

is it clear now?

sarcastic much?

Oro

sarcastic much?

yes or no?

no

Oro

I did see this in the issue report, so may be worth copy-pasting over directly:

Add -D legacy-wayland=bind-wayland-display to the mesa debloat scripts and rebuild the packages/AppImages. Restores the old fallback; kmsro is already built for the aarch64 mesa (zink pulls it in on 26.1+, the ARM gallium drivers before that). Upstream: fall back to device->nodes[DRM_NODE_PRIMARY] in the feedback handler when there is no render node. No MR for this yet (!38079 only touches loader_get_user_preferred_fd). Workaround: WLR_RENDER_DRM_DEVICE=/dev/dri/renderD128 so the compositor's main_device has a render node.

(edited)

^

12:38 AM

mentioning this in some way, as was found from testing, would be actually useful

Oro

mentioning this in some way, as was found from testing, would be actually useful

we didn't test WLR_RENDER_DRM_DEVICE=/dev/dri/renderD128 though

sounds pretty confident in https://github.com/pkgforge-dev/Anylinux-AppImages/issues/927#issuecomment-5878923053

Oro

sounds pretty confident in https://github.com/pkgforge-dev/Anylinux-AppImages/issues/927#issuecomment-5878923053

yes but that is an LLM output I wouldn't assert with such confidence in the mesa bug tracker all I can say is that eglgears_wayland works and doesn't work with xyz combination

that was not disclosed as LLM output

Oro

that was not disclosed as LLM output

nemo is a bot xd

....joyful.

12:41 AM

I'm dropping this topic, since I don't think I can really trust that any debugging done here is actually useful or in good faith.

1

Oro

I'm dropping this topic, since I don't think I can really trust that any debugging done here is actually useful or in good faith.

are you sure? because assem11assem tests do show a problem very clearly

I never said that it wasn't a problem, but that I can't trust that any debugging done, or any possible solutions listed, are actually solutions or verified.

Samueru

is it clear now?

@Oro how can you not trust this terminal output?

12:43 AM

this is native archliunuxarm with native mesa and packages

12:43 AM

it shows the problem very clearly

I never said that there wasn't a problem, but that provides so little information that it might as well not be included

if you say so

Oro

I never said that it wasn't a problem, but that I can't trust that any debugging done, or any possible solutions listed, are actually solutions or verified.

and the problem was gone after I added -D legacy-wayland=bind-wayland-display btw

which is not a proper solution, and it was laid out exactly why that flag is legacy now.

12:46 AM

This extension was originally created in a pre-dmabuf world, where we didn't have a universally-accepted way of exchanging buffers between client and compositor, or even really the ability to describe formats and modifiers universally. Since then, the world has settled on dmabuf with DRM FourCC and modifiers. We've had the zwp_linux_dmabuf_v1 protocol for 10 years now: both clients and compositors implement this protocol to handle buffer sharing. Compositors either use EGL_EXT_image_dma_buf_import or the Vulkan dmabuf extensions to import these into GPU world.

Oro

which is not a proper solution, and it was laid out exactly why that flag is legacy now.

so can you confirm eglgears_wayland is has never been fixed here?

12:47 AM

and Qt6?

reword that

Oro

This extension was originally created in a pre-dmabuf world, where we didn't have a universally-accepted way of exchanging buffers between client and compositor, or even really the ability to describe formats and modifiers universally. Since then, the world has settled on dmabuf with DRM FourCC and modifiers. We've had the zwp_linux_dmabuf_v1 protocol for 10 years now: both clients and compositors implement this protocol to handle buffer sharing. Compositors either use EGL_EXT_image_dma_buf_import or the Vulkan dmabuf extensions to import these into GPU world.

this is why I asked this

12:48 AM

because I don't think Mesa devs intended to break all of Qt

I am saying that enabling those legacy flags is not the proper solution, and that even if apps are broken, there is a proper fix somewhere else in Mesa

12:48 AM

why do you need me to confirm that Mesa did or did not intend to break QT or eglgears?

12:48 AM

what bad faith question is that?

Oro

why do you need me to confirm that Mesa did or did not intend to break QT or eglgears?

because it seems crazy to me that they would break all wayland clients that suddenly

it wasn't suddenly, it was over a year ago

including their own demos

I don't know if this is clear to you, but this is a bug and deserves to be solved, but how you are going about this issue is not the way to solve it long-term.

12:50 AM

And blaming this all on the deprecation of some EGL extensions isn't going to fix your problems.

Oro

I don't know if this is clear to you, but this is a bug and deserves to be solved, but how you are going about this issue is not the way to solve it long-term.

I cannot ask if they intended to break all those apps???

The MR you linked has the rationale explained, as well as the way forward for any app still using that extension;

12:51 AM

the MR linked has the rationale for hiding the extensions, and what apps should do in the absence of those extensions; if something is broken, that is a bug

12:51 AM

how much clearer do I need to be?

I think the greater issue is that logistics in FOSS sucks.

Oro

The MR you linked has the rationale explained, as well as the way forward for any app still using that extension;

yeah that's crazy if they actually broke eglgears_walyand and qt6 on purpose but I think what happened is that the mesa devs didn't realize the extend of affected clients by that change, this is why I asked them if they intended to break Qt6 and their own demos.

if it was intended, it would've been documented somewhere.

12:52 AM

And your bug would have been closed.

12:53 AM

If this is a case of clients breaking because they're relying on outdated extensions, then this is a client bug.

lizbeth86

I think the greater issue is that logistics in FOSS sucks.

walyand sucks*

Samueru

walyand sucks*

I mean

which one? :)

Last time I used wayland on my machine it just gobbled up 100% of the CPU and I could barely do anything

12:54 AM

Of course I could report a bug for that.... but I'm on FreeBSD, nobody cares

where would you report that bug to? 🫠

12:54 AM

surely not libwayland.

lizbeth86

Of course I could report a bug for that.... but I'm on FreeBSD, nobody cares

the only people I've noticed that actually care about wayland bug reports are the SDL devs xd nobody else really cares, a while back there was some massive regression in mesa related to GTK4 and intel igpus that took 8 months to be fixed

12:55 AM

in wayland

12:55 AM

@Oro enjoy https://github.com/pkgforge-dev/Anylinux-AppImages/blob/main/HALL-OF-FAME.md#garbage---wayland And I need to update this include more recent issues we've run into like the gamescope regression

Oh yeah gamescope just straight up crashes for seemingly no good reason

Samueru

@Oro enjoy https://github.com/pkgforge-dev/Anylinux-AppImages/blob/main/HALL-OF-FAME.md#garbage---wayland And I need to update this include more recent issues we've run into like the gamescope regression

so you're blaming... wayland? as a protocol? for gnome's problems?

meanwhile SDL, fixed ASAP: https://github.com/libsdl-org/SDL/issues/16560

I mean, I can go ahead and implement wayland on top of X11 if that floats your boat.

12:57 AM

then bam, wayland fixed, life good.

Oro

so you're blaming... wayland? as a protocol? for gnome's problems?

what's crazy is that the Wayland issues are problems that should not affect us, they are problems for the implemenations, clients and drivers, yet Wayland is so broken that we often waste time at bug reports because there is no way an application can be having this issue and it turns out that is true!

To be fair; wayland would be 100% better if wayland people forced everyone to use wlroots instead of making 48593458 fragmented impls

breaking news: protocol implementations cause bugs, bugs cause bug reports

12:58 AM

life continues as normal

lizbeth86

To be fair; wayland would be 100% better if wayland people forced everyone to use wlroots instead of making 48593458 fragmented impls

not really, I wrote about this xd I used to think that Wayland would be fixed if everyone used a common library like wlroots, but guess what 😹 Some wlroots compositor was crashing while sway didn't

Samueru

not really, I wrote about this xd I used to think that Wayland would be fixed if everyone used a common library like wlroots, but guess what 😹 Some wlroots compositor was crashing while sway didn't

Ok i should add, using wlroots AND not being outdated as shi

12:59 AM

I mean even solaris has the same X11 headers as my FBSD install

I genuinely don't know what to say here other than "sucks to be you, but I don't have any problems"

works on my machine tm

file a bug report and move on, because that's how people become aware of bugs and can look at fixing them

1:00 AM

instead of blaming a protocol and degrading the user experience to make your egos happy

Oro

instead of blaming a protocol and degrading the user experience to make your egos happy

it's never fault of the protocol

I dunno, you seem to be really quick to blame wayland itself too

1:01 AM

and then blame an implementation for all of wayland's problems

The protocol is fine, the implementation is dissapointing.

lizbeth86

The protocol is fine, the implementation is dissapointing.

the protocol is not fine, recently there was another drama about adding a hotkey extension because people realized using portals for this sucked

the Intel issue you linked in regards to gtk4 in the list of

wayland issues

also includes this:

The problem occur in Wayland and x11

Samueru

the protocol is not fine, recently there was another drama about adding a hotkey extension because people realized using portals for this sucked

I mean I think the dmabuf thing is pretty nice

otherwise on UX terms... ehhh

1:03 AM

I prefer my X-utils for keyboard/input

Oro

the Intel issue you linked in regards to gtk4 in the list of

wayland issues

also includes this:

The problem occur in Wayland and x11

well when I had the issue reported it was fixed by forcing xwayland but I believe that is due to gnome defaulting to opengl instead vulkan in x11 iirc.

1:10 AM

@Oro We actually compile gtk4 without the vulkan backend, because sometimes the variable is not enough.

1:11 AM

I had to use remote into the person's computer to discover this, some ancient laptop with intel igpu, even the native arch gtk4 was broken xd

>stop complaining, just file a bug report and move on! Said after complaining nonstop instead of filing bug reports and moving on

lol

crueter

>stop complaining, just file a bug report and move on! Said after complaining nonstop instead of filing bug reports and moving on

@Oro Say hi to the hall of fame

https://github.com/pkgforge-dev/Anylinux-AppImages/pull/986

add gnome incident to the hall of fame by Samueru-sama · Pull Requ...

Did you ever ~~get~~ grow out of your high school habits? (edited)

3:49 AM

I can't tell.

3:50 AM

This is going in the history books right next to probono.

Wait you're a GNOME developer?

3:51 AM

LMAOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Ah, fuck off.

3:52 AM

All yall are doing is making yourselves look bad.

Oro

This is going in the history books right next to probono.

probono didn't bother to report and fix wayland bugs At least I tried

If "ooo you're a gnome developer" is the worst insult you can pull, I fear for your emotional maturity.

3:55 AM

I've got cousins who can do better than that.

3:55 AM

At least be creative and make me laugh.

Well it seemed to work

4:02 AM

I don't know how you expect this to look bad on me.

4:02 AM

It's really just a highlight of how bad yall are at not slinging slander around.

Looking through your message history here, it seems like you've done exactly two things:

Joined the server specifically to complain about AppImage and shill for Flatpak (while being extraordinarily hostile to everyone who disagreed with you, mind you)
Did absolutely nothing else until you came to complain about us forcing X11 (which, again, is due to Wayland being broken)... then got hostile yet again (while being a raging hypocrite on top of it all)... then complained about people returning the favor Safe to say this is a fairly justified reaction. Which wouldn't be an issue if you weren't such an incessant hypocrite who can't handle even the slightest bit of dissent, let alone the fact you have contributed nothing of value and joined this server specifically to complain over and over again

6

I don't see a single mention of "the flat that must not be named" that you say I'm shilling for.

4:12 AM

(stupid moderation filter. Wow.)

4:13 AM

Yall were the ones who wanted to make this conversation go this way

i dunno....

Thats, five months ago?

4:13 AM

How is that relevant now?

crueter

Looking through your message history here, it seems like you've done exactly two things:

Joined the server specifically to complain about AppImage and shill for Flatpak (while being extraordinarily hostile to everyone who disagreed with you, mind you)
Did absolutely nothing else until you came to complain about us forcing X11 (which, again, is due to Wayland being broken)... then got hostile yet again (while being a raging hypocrite on top of it all)... then complained about people returning the favor Safe to say this is a fairly justified reaction. Which wouldn't be an issue if you weren't such an incessant hypocrite who can't handle even the slightest bit of dissent, let alone the fact you have contributed nothing of value and joined this server specifically to complain over and over again

Re-read this message carefully

My purpose in this conversation here was to make sure eden handles display servers properly, find some information on why eden insists on displaying warnings for Wayland (which, one aarch64 issue, and then I don't recall any other linked bugs for that in regards to performance or other crashes in this entire conversation), and get the fuck out of here.

4:16 AM

Hell, you'll notice earlier that I even noted that making eden just set the QT platform to xcb;Wayland was perfectly fine with me and exactly what I wanted.

Oro

All yall are doing is making yourselves look bad.

sorry your totally right and the devs should get right on it

And it still keeps your funny reasons for wanting X11 without screwing over anybody else.

4:16 AM

So.

4:16 AM

Not quite sure what I did wrong here besides pointing out very bad communication on yalls part.

4:17 AM

If you can call that wrong. Somebody got their ego bruised.

4:18 AM

You can try to bring up conversations from five months ago and they hold no weight for anything here, because "the flat that must not be named" was not the focus of any of this conversation.

4:19 AM

Yall are really just being immature and trying to harm people for idiotic reasons

well, if you contributed anything of value to this server other than complaining it wouldn't be brought up in the first place...

I didn't lose my temper, lash out at anybody, or spread slander

4:19 AM

(guess who did?)

crueter

well, if you contributed anything of value to this server other than complaining it wouldn't be brought up in the first place...

Hey so this is your problem, not mine :)

4:20 AM

You're the one wanting to bring this up.

4:22 AM

I'm not quite sure what you're looking to get out of this.

I derive enjoyment from your incessant whining

crueter

I derive enjoyment from your incessant whining

crueter your actually the problem and you need to fix wayland on eden right this instant (edited)

That reminded me of when I got shell shocked last week because the Ubuntu VMs my workplace uses use GNOME, and there isn't a built in clipboard history feature (which even macOS has now)

crueter

That reminded me of when I got shell shocked last week because the Ubuntu VMs my workplace uses use GNOME, and there isn't a built in clipboard history feature (which even macOS has now)

crueter

That reminded me of when I got shell shocked last week because the Ubuntu VMs my workplace uses use GNOME, and there isn't a built in clipboard history feature (which even macOS has now)

Womp womp

man the gnome dev stereotype really is real huh

1

The jokes write themselves

"why isn't there (feature thats on every other de and os)" "because we don't like it fuck you"

It's not nice to insult yourself, crueter.

Oro

It's not nice to insult yourself, crueter.

can you add a maximize and minimize buttom to gnome pretty please, i promise i'll use it if you do

1

I'll use GNOME if a clipboard history feature gets added