The Wayland Protocol To End All Wayland Protocols
Vložit
- čas přidán 24. 01. 2024
- I've discussed plenty of controversial Wayland protocols but honestly who would have thought that just having the ability to change the icon for a window would be this controversial
==========Support The Channel==========
► Patreon: brodierobertson.xyz/patreon
► Paypal: brodierobertson.xyz/paypal
► Liberapay: brodierobertson.xyz/liberapay
► Amazon USA: brodierobertson.xyz/amazonusa
==========Resources==========
Wayland Icons Protocol: gitlab.freedesktop.org/waylan...
Old Proposal: gitlab.freedesktop.org/waylan...
=========Video Platforms==========
🎥 Odysee: brodierobertson.xyz/odysee
🎥 Podcast: techovertea.xyz/youtube
🎮 Gaming: brodierobertson.xyz/gaming
==========Social Media==========
🎤 Discord: brodierobertson.xyz/discord
🐦 Twitter: brodierobertson.xyz/twitter
🌐 Mastodon: brodierobertson.xyz/mastodon
🖥️ GitHub: brodierobertson.xyz/github
==========Credits==========
🎨 Channel Art:
Profile Picture:
/ supercozman_draws
🎵 Ending music
Track: Debris & Jonth - Game Time [NCS Release]
Music provided by NoCopyrightSounds.
Watch: • Debris & Jonth - Game ...
Free Download / Stream: ncs.io/GameTime
#Wayland #Linux #OpenSource #FOSS
DISCLOSURE: Wherever possible I use referral links, which means if you click one of the links in this video or description and make a purchase I may receive a small commission or other compensation. - Věda a technologie
At this point Wayland maintainers must be sweating whenever they see a new PR by Matthias.
nah they save their anxiety for the brodie video that follows
I didn't know Linux had PvP
In linux world everything is PvP
More like DvD (Developer vs Developer)
@@willyantowilly7165 DnD (Developers nagging Developers)
Wayland is the best sitcom I've seen in years.
Sometimes I get really tired of the bureaucracy at my workplace. But holy crap, I have it lucky compared to wayland proposal authors.
Your workplace probably has a boss, this is what happens when there is no boss and only engineers who disagree with each other
@@BrodieRobertsonI was about to say that it's high time that somebody took the leadership and just decide what is being done, instead of an endless number of quasi-religious lunatics. I'm a big fan of Qt/Kubuntu/Plasma, but at this point I hope that System 76 will succeed with their Cosmic DE in such a way that they can do whatever they want with Wayland and all others have to follow. Everything's better than this travesty. This is just ridiculous.
@@stephanhuebner4931 I can't speak for certain but from what I've experienced with people like Jeremy Soller, mmstick and a few other System76 people they seem more akin to the KDE people as in looking for shared solutions not just something that works for them
@@BrodieRobertson It wouldn't need to be a solution that only works for them. It would be enough if they'd get the majority of devs/DEs on the same page, up to a point where Gnome had no other chance than to follow or become irrelevant.
Halting decision making until everybody agrees on one solution works in theory only, as is proven by the Wayland-development process time and time again.
@@BrodieRobertson sure, but at this point I have the feeling that Wayland would move not just faster but also better if KDE, Cosmic and wlroots make their own separate repository where they create (and implement) a set of extensions
that bloke swearing he will guarantee a NACK is so pathetic lol.
Yeah, that was just ludicrous. How can you be so resentful? It's no surprise that people think GNOME developers are a bunch of pricks when hostile people like Sebastian are upfront on these discussions.
If GNOME keeps nacking things, maybe THEY should be nacked.
Needs a "be excellent to each other" COC, then ban his ass.
Mattias Klumpp, the man who talks to Wayliens.
a 28-minute video on a discussion about setting window icons... i love this stuff.
real this is amazing
to be fair the only one complaining there was a gnome developer
linux/foss world in a nutshell.
@@vilian9185 That's par for the course. The repo would probably be much more effective without gnome inclusion. I'm sure it's not actually right, but way too often gnome actors _seem to_ be just obstructionist without any good arguments.
i heard something about microsoft giving a bunch of money to gnome and now i'm finally starting to understand how that would benefit them... the more money gnome has, the more they'll be a pain in the ass and try to set everything an fire, resulting in more people going back to windows.
For when people say, "Just recommend a protocol" to get around Wayland's shortcomings
Recommending a protocol is easy getting it merged is hard but it has to be done
Yeah, create a hidden file every time you want to change window icon, another banger from wayland team
Which is so strange considering how minimal Wayland is supposed to be. I mean, I already have to set a bitmap in order to have a mouse show on the window at all, almost everything else is done through a surface, but not window icons? You set the window title through code, you set the class through code, why not those in the desktop entry as well if app imitation is such a concern?
If nothing else, it's inconsistent.
Does this mean I'll be able to play DOOM in the window icon? That's really the most important thing.
That's the real question
Someone is asking the right questions.
I ran into this madness with Wayland about a week ago and thought "huh, that's weird, I just want to set an icon". Then I see this video and I'm absolutely shocked that Wayland managed to come into existence at all, given this insane amount of debate about absolute bullshit. If they just did the simplest thing instead of arguing about nothing they'd have solved computer science by now.
the main reason why they came to be at all, was that it was only 2-3 people actually specifying anything originally
and they weren't the ones who came up with xdg-toplevel
Matthias: *type "print("Hello, world!")"
Wayland Devs: *starts grabbing pitch forks
Wayland Maintainers: *sweating profusely
phew, i was beginning to worry we wouldn't get a few big Wayland threads this year :)
And while it's ridiculous, i'm thankful to people like Matthias who are actively working on it to make some progress. See you all in half a year probably for the update.
I feel like one of the most common use cases are like, browser profiles, or apps that display status on the icon ala skype
This is already possible without changing the Icon
what about individual windows for each conversation, showing persons avatar as an icon?
It really feels like some of these wayland devs are disagreeing just for the sake of disagreement, it's a bit exhausting to read sometimes
It's not for the sake of disagreement, it's because they really like .desktop files that much. They've got a vision that everything that runs on your computer should have a .desktop file no matter what.
*GNOME* devs are disagreeing for the sake of disagreeing. Most of the other developers seem to get along somewhat enough to have useful conversation even if they don't always agree, but when GNOME cuts in they always just say, "this is useless, we don't need it" and shut it down because they'd rather their toy of a desktop environment be *literally* broken to fit their own vision of the desktop than implement core functionality that everyone else wants.
On the BSD side there is a term for this kind of bickering over minor implementation details: bikeshedding. (Parkinson's law of triviality)
Another quality and informative video, Mr. Wayland
Passing the wl_surface seems totally reasonable. Surfaces can already be displayed multiple times, they can already be transparent to look nice on different backgrounds, because the window itself is a surface. This would let us do animated icons, etc. Essentially the icon would become a second, smaller window, which I find kinda nice.
Can they be resized?
Also I guess transparency or not, they don't want to draw the light mode icon on a dark mode background.
@@Poldovico AFAICT the proposal says the compositor would decide how big the surface is. So it could pick the biggest possible one (or biggest actually on screen), then scale down for smaller ones. For the light/dark mode icons, I think many apps only have the one icon as it is, i don't see how this is worse.
Honestly, I just want to see Xeyes in a Wayland icon.
@@stribika0 Yeah, I don't buy the dark mode thing either, but it seems like they're hellbent on it.
@@Poldovico Need light mode, dark mode, color vision impaired modes 3 through 7, short attention span mode, cubist mode and waifu mode. Your application must have unique icon art for each or we will ban it as a security threat.
I think wayland would be a lot quicker in development/adoption if the potential protocols were implemented first (maybe marked experimental) so compositors have a chance to try it and see what problems actually need solving, and users who urgently need the protocol can have the functionality, even if it is subject to change
The C++ committee requires 2 implementations in different compilers before accepting a change to show that'l it's even possible and to find out if there are things in the standard that should be revised to aid in implementation. It works very well and should be the defacto way of creating standards
@@__Brandon__ the Wayland equivalent would be two compositors implement a protocol extension, and work out the quirks on their end before even bringing it up with the Wayland project at large.
@@__Brandon__ I like the ECMAscript method (similar to that), they start with a proposal (stage 1), then refine it to stage 2 and mock up some test implementations, then they advance to stage 3 and browsers can implement it. Once two have done so and it works it goes to stage 4 and becomes standard
@@jfolz it would be nice if they had 2 implementations before starting the bike shedding. It would at least keep the bike shedding targeted
@@__Brandon__ it would definitely help. Xaver Hugl (the KDE dev) took issue with the fact that, as defined, setting the icon is an asynchronous process. Since it can take some time for apps to reply to icon requests, they would either need to delay showing the window, or the icon may flicker. Personally, I don't think this would be an issue, as we're talking a few milliseconds of response time. That's _ages_ in computer land. If you had working implementations to show that it's not an issue in practice, you'd have a much easier time to convince people.
I wonder how wayland is going to look in 10 years, once all these things are sorted out. Hopefully all ends well.
same way it looks now, it's already had 10 years to sort all these things out, and 5 years for initial development
like Xorg.
@@thewhitefalcon8539 Could be. To me it seems like wayland is finally starting to be useable for many people so I'm quite optimistic that most of the issues that keep people on x11 will be fixed in the next 10 years.
@@thewhitefalcon8539 the biggest and quickest improvements came in the last 4/5 years, so it's going to be totally different now. In the last years, Wayland has progressed immensely.
IN GOD DAMM 10 YEARS!? 💀😵
I want *my* window icons to be dynamically generated, to show a progress bar, animated emoji, or a whole movie in 16x16px if I wanted to!
same
I want maximise to be an animated image of the peanut butter jelly time banana
Ah, you mean like the MacBook keyboard with function bar and the ArtLebedev keyboard where every key has it's own full colour LCD screen?
The LibrePCB icons (an educated guess). The first is "component view" ie. where you pick the individual components like resistors, capacitors, voltage regulators and MCU/SoC and so on.
The second is the 'PCB view', where you place the components on the PCB.
And the third one is "routing view' where you layout the pcb traces (the connections between components. :)
I thought that it was components, traces, and schematic but I haven't used LibrePCB so just basing it on alternatives I have used.
@@TsudicoI think you are saying the same thing in different words.
If it's anything like Eagle, Kicad and every other electronics CAD package ever, it's probably a library editor, a schematic editor and a PCB editor.
a true spectacle of the brodele library
XDXDXDXDXDXD
What I don't understand is why. From a lot of these discussions it just seems like it's only GNOME and GNOME developers that are doing stuff like this, for the most part a lot of the other DE contributors were bringing up issues that are valid, and those issues were properly addressed and thus those DE contributors were content with the new protocol, for whatever reason GNOME always either makes it a "our software center" or "Server-Side Decoration" issue, and in this particular discussion it was literally a full bingo. Saying that A. We need .desktop icons for the software center B. that this shouldn't be added because it's"Server-Side" decoration and C. "these use cases are made up, and isn't matching the use cases that GNOME supports so bad".
I do understand that discussions for things like this are going to take some time, it's protocols that reach feature parity with X11 while doing it in a fundamentally more secure and performant way, and doing stuff like that will have a lot of inherit issues trying to get it supported for all implementations, but intentionally saying that we shouldn't add this X11 feature at all because then you just have X11 again is missing the point entirely of Wayland, and Open Source. Reach feature parity with X11 (which XWayland fills the gap of for the most part) then do it a different way later on.
At this point I'm not sure why anyone is even listening to GNOME at all, because either they'll slow down the development of Wayland, or when a new protocol is added to Wayland they just remove it when compiling it for their DE anyways. Again, are there going to be growing pains yes, but the idea is to get it added, see what issues need to be fixed (that's what betas and testing is for) and fix them. Don't just shoot it down without providing a more substantial reason than "I don't like it" and even without seeing how it would work in practice.
Yeah, even if it splits Linux, the other Wayland developers should kick out the GNOME devs from making decisions.
GNOME really should be kicked to the curb. They've held the development of the Linux desktop back for years because of their own selfish and petty nonsense.
They are listening to gnome because gnome has had the advantage of being the default desktop in all major distros since always. Hopefully now that plasma has the same release cycle they'll either have to catch up or be abandoned, rightfully so
In this episode of Why Wayland Is Awesome, Brodie finally discovers the reason why so many other people think Wayland is not awesome.
I am possibly the most critical of Wayland's development I have 3 videos and a 2 hour interview on the multi window stuff, but I'm also hopeful for the direction of the project I just wish we travelled in that direction a bit faster
This is why we can't have nice things. Same as with old-time Linux users who don't want to change things to become more user friendly because it would make "their" OS less exclusive and special ... And know how to do it anyway or don't need it themselves - edge cases or different points of view aren't a thing to them ... This is why a lot of people who could won't contribute - to keep their sanity.
Old timey Linux user here. IDGAF as long as it works, & don't stray far from commom sense like Wayland does. I'm all for progress & change, but this Wayland fiasco is unnecessarily, convolutedly complicated.
Wouldn't be a problem if they don't pitch a fit about simple stuff but go out of the way to make the complex parts work.
The thing that infuriates me about the people who fight against making Linux more user friendly - is that these are the same people that also say that Microsoft and Apple are evil and the market needs an alternative to get rid of them.
Guess what buddy, that alternative is Linux, and if you want normies using it, you need to make it easier for them.
I see the title before clicking on it, and my immediate thought is "Ok, what did Gnome do this time?"
They did everything ma' boy... They did everything they could to ruin everything there is...
- Linux has a lot of user-facing transient programs that only exist while they're running
- Learning from Xorg, there's a major push in Wayland to move things into xdg-desktop and out of the compositor
- xdg-desktop is data-driven and doesn't associate applications with processes unless strictly necessary such as for IPC and GUI.
This is clearly a conflict, one of these will clearly have to suffer. I would really rather it weren't transient programs, as virtual environments of all kinds are one of the best features in Linux.
Maybe xdg-desktop isn't the solution for everything...
@@ScibbieGames xdg-desktop manages the data we want to extend. The question is whether we want to empower xdg-desktop to recognize temporary data, or bypass it and have consumers talk to transient sources directly. I think keeping all the data unified is useful, but at the same time consumers that aren't designed with changing data in mind almost always break when the data does change.
If an app disguising itself as another app is really such a security concern, why don't desktop environments just support showing both the store-approved icon _and_ the app-requested icon at the same time when they're different? This way there's no mistaking which app the window is from, and different windows from the same app can still be distinguished. On Windows, app icons can have "badges" which the user can even turn on or off in the system settings, this is used already for distinguishing different browser profiles in Chrome for example.
Yeah, this is what makes most sense in my mind - just superimpose a slightly smaller and offset icon defined by the window on top of the "master" icon. If the protocol was being handled by the compositor, they can do whatever they want with the "window" icon anyways.
Also, if you use suchspecificly malicious app, it is your problem
@@mckendrick7672 exactly! If you have multiple Chrome profiles on your windows taskbar they each have the Chrome icon as their main, and the profile picture overlaid at the bottom right. It would be great to have the same in Linux, and would work for all multi-window apps.
@@ukyoize These malicious apps probably wouldn't be used intentionally. Imagine you have an app which appears to be normal but it also has a browser inside it. It could open links in the app using it's browser and disguise it as official firefox (for example). This would allow stealing passwords the user enters and things like that. They could even mimic a keyring popup to try and steal your root password (although that's DE dependent). This could happen even if the app was perfectly sandboxed and has minimal permissions. You could blame users for that, but many apps are closed-source or niche anyway, so problems with them might take a while to surface. Social engineering is the biggest security risk as always.
This is not an impossible problem though, if the DE clearly gives some indication that a window is from a particular app that pretty much solves it. Then it would be more obviously malicious, at least to enough users that the app would quickly get a bad reputation and would be taken down if possible.
As a Windows and WSL user, oh my God what the fuck.
I had like a whole spiel about how I support open source operating systems and such, but honestly, just what the fuck.
@@allideni836yeah as a windows user seeing how this whole thing is managed its ridiculous, like why can’t people just be happy letting you just dynamically change the icon without modifying the file system ???
Textbook definition of design by committee. The entire debate regarding "but what about all the other desktop file features" was pointless. Just let the application set a custom icon per window, why is that such a foreign concept? What happened to KISS? The mere idea of messing with .desktop files to hack this in should be laughed out of the room, yet here we are. The leftpad-generation is in charge now.
What I really don't understand is that the people talking about using .desktop files for this seemed to be saying it was a simpler solution? To me it sounds convoluted to use a .desktop file just to set the icon for a window. Do you also need to save a .desktop file to change your window's titlebar text?
I know! It's fucking insane, right?! I just added follow notification for this because idk it seemed like it'd solve an issue I notice with those ugly W icons. I never expected this to go all this long with so much drama. On the one hand, it's good drama, on the other hand, I just want my goddamn icons already man.
Wayland devs trying to render a picture (IMPOSSIBLE)
I saw some FreeBSD users talk about creating their own display server because of crap like this.
I tried replying a link but youtube keeps shadow banning me
If you happen to know where let me know
A big plus of Wayland is that you can implement whatever you want, regardless of it being "accepted" or not. If stuff like this is a concern, BSD devs could make "Wayland BSD" protocols that a client is able to opt-in to.
Then you get the advantage of not having to deal with ridiculous politics and still get compatibility with existing Wayland projects.
Pinnacle of Open Source software development.
I would love for you to interview probonopd on Wayland and such.
I mean, the term 'Bikeshedding' is literally about everyone having an opinion about the color we paint the bikeshead (something purely cosmetic)... so yeah... it should be obvious that there's going to be much argument about anything cosmetic... and with wayland EVERYTHING is cosmetic.
Valve needs to grab them by the balls (or lack therof) and create a specific development path for Wayland.
Funny you bring up Valve because in the documentation for VR if you're on Wayland and Gnome the suggested fix is just install KDE
I could see them go at some point "here, we created this protocol, Gamescope uses it, there you go; whichever environment doesn't support it, their problem"
@@BrodieRobertson sounds to me like a good fix, I approve.
@@kuhluhOG That's Valve's if-everything-else-fails nuclear option.
@@Megalomaniakaal sure, that's why I said that I could see them doing this at some point
PWA's in chromium based browsers suffer from this heavily.
I'm genuinely considering making my own X11 alternative.
With the amount of stuff going on to make Wayland development slow, I might even have a chance of catching up to them!
You could just fork x11 rather than starting from scratch.
@@Megalomaniakaal If half of what I've heard is true, then that would be a bad idea. It would be easier to start from scratch than inherit the piles of technical debt and atrocious code contained within X11.
@@bluesillybeard but that's how we got Wayland!
@@bluesillybeardmost of the X11 code isn’t that bad. The problem is GTK wants to bypass 90% of it. QT generally goes along with it because they don’t want to maintain an X11 port and a Windows port. The driver initialization code needs to entirely ported to KMS at this point which would break machines that don’t have a KMS capable FB driver. But most of those cards are pre PCIe and amd64.
If you ever do start an X11 replacement project, remember to start with a generic communication protocol first. Note, not an _extendible_ one, a generic one- you want to be able to send arbitrary command-wrapped data down it for new commands without _ever_ rewriting it. Then try to make it 7-bit compatible, just in case, and you'll have a good base to build on.
I have nothing, this is business as usual. And before anyone says it's unique to Wayland or OSS, it's actually worse in business, you just don't hear about it unless it's so bad that someone ends up dead or the whole company goes under.
The good thing about OSS is you can mostly just ignore what they're doing. Wayland sucks, so just run Xorg and you can ignore it.
@@thewhitefalcon8539 except you can't, at least unless you are going to switch your desktop over to a rootful XWayland session. I'm a long time user, on several types of systems. X has been fraying at the edges since the 90's and was fully broken in some ways by the 2000's. Using vendor hacks to get working systems with irritating caveats has been life before Wayland. Do I think the design is fantastic? No, but it isn't garbage and it's not going anywhere.
You know that on a system with compatible drivers Wayland is lighter than X right? Corporations who are used to rolling their own UI frameworks are already switching to Weston and wlroots based things already for embedded work. Scientific applications with problems nearly all work fine in XWayland and things are rapidly improving there.
No matter what you say about Wayland's incompleteness, it's what we have. The instant Nvidia stops pretending Wayland doesn't exist XOrg will be buried on Linux outside (essentially) emulation, hardware archivists, and retro hobbyists. It's the end of an era and you can't tell me actually lose anything by going along with it.
This is worse because it's a headless chicken flailing around. At least in business there's someone in charge. Typically a complete idiot, but at least things progress instead of being stuck like this. Imagine telling your boss you spent weeks arguing over how to display an icon.
@@bleack8701 You can't fool me, I've literally been at those meetings. It happens in business all the time.
@@bleack8701 You know this literally happens in business all the time right?
If more of the Wayland developers had the mentality of the guy proposing this PR, and less of them had the mentality of Sebastian and the others, Wayland probably would have been done a long time ago and been good enough for almost all the X11 holdouts... and I say this as someone who likes Xorg. It's pretty plain to see that the problem with Wayland is that there are too many people refusing to implement reasonable features because of pseudo-arguments about "security," and this is precisely why a lot of projects choose to stick with Xorg instead... because it just works and lets you do things while Wayland is all "No, you can't have this," and puts the binders on. I feel like that weren't the case, Wayland would be far more popular, but instead we just have the majority of people who are happy with their Xorg environments seeing Wayland as broken because it won't implement stuff they like having and which is also available on Mac and Windows.
One thing I really think was a bit short-sighted on the part of Wayland developers is that they thought that by creating this whole new thing rather than improving X11, they could just shut out the voices that resulted in a lot of "cruft" being kept in X11. What they didn't understand is that now that there's talk of deprecating Xorg, all those voices that didn't have a seat at the table the past 10 years while Wayland was being designed are going to be coming and asking for the exact kind of features they didn't want to implement. Everyone was willing to tolerate Wayland being a toy for people who want to experiment with new designs and don't care about little details like this, until they were told Wayland was going to be required someday... and now, those same voices that the Wayland devs were probably hoping to avoid dealing with by starting Wayland and putting X11 on life support are going to be coming back and making their voices heard, saying the exact things they clearly didn't want to hear.
Wayland developers don't want to be stuck for another 30 years with software designs they hate, much like what happened in X11 (which has a POS codebase that no one wants to touch), that's why they are so picky with the protocol designs. The X11 paradigm is not compatible with the Wayland paradigm in many ways, and this is the only shot (as in, the chance to make breaking changes) to get it right.
X11 works fine in many cases because it just allows anyone to do whatever they want in the system, which is ridiculous. This is why keyloggers are so easy to implement in X.
I get just as furious when I see so much bikeshedding in protocol discussions, but I do understand, as a fellow software developer, they are not some sort of evil entity that wants to ruin everyone's day. They are just developers that don't want to hate what they are working on.
Nah, you're way too optimistic. Consider the actual reason X didn't change after a certain point. That's because both OSS and proprietary devs were unable to build something everyone would be able to use, and it was usually intentional.
@@lucas7061 Well, I just see so many cases these days of designs that take a lot of freedom away and create something that feels like a prison/asylum instead, where a lot of things you want are denied you because the guards don't trust you. That said, I'm sure there is some kind of middle ground that can be found, but honestly the way things are going, I get the feeling what's actually going to happen is the Wayland standard will be fairly spartan, and all the stuff people actually want will be implemented in different, incompatible ways by different compositors. No one is going to settle for the base Wayland protocol for long, and it already seems like people are working around the problems with Wayland by doing things using other libraries and protocols that are separate... sort of like the Qt/GTK divide but even bigger now. Red Hat has even created a libei, for instance, to work around Wayland, and I suspect others will do the same. So now instead of one big X11, we have Wayland, along with other protocols that do what Wayland won't, that all have to be baked into the application, and which work with varying degrees of success on different compositors. I get the feeling a lot of work is happening just to reinvent the wheel in the end, but as long as we do get the wheel in the end and not a rectangle, I'm happy.
Brodie usually takes the side of people like Sebastian btw.
@@lucas7061 There's no reason the X11 protocol has to be tied to the Xorg codebase. If nothing else, someone can fork Xorg and strip out all the stuff they don't like.
Wow, wayland is going to be a great learning xp for the third windowing system. "Second edition is a rough draft for third edition."
Third system will probably be some extensions to X11
Linux community tries to implement setting a window icon without starting a fight (impossible)
How is this 28 minutes long oh noooo.... let's watch...
IMO some of the functionality that was present in X11, but isn't in Wayland should be provided by xdg-desktop-portal or some kind of user service, such that the core functionality only needs to be coded once, and all compositors/window managers/task managers would have to do is the plumbing for some basic IPC with that service.
It could also be specified that the service handles scaling to the requested size, that each icon is given a UUID, allowing users to override/replace icons, or that the service stores requested window positions for multi-window applications aswell.
I 'm probably over-simplifying this but in principle, if each Window is already represented by some kind of object with properties, and if there is already a convention for how to set a title property for each Window, then why can the approach for setting the icon property for each Window not follow the same convention?
I think the part that is nearly always valid is to check whether any particular proposed feature should live inside the compositor implementation or belongs (is better modelled under) something else. But even that should be some simple criteria on a checklist rather than generating pages of discussion in its own right.
I ask because it is already a well-established pattern in application programming that
an application ≠ a window
(as in not a 1 to 1 thing) an application may or may not happen to display one or more windows or dialogs but the window is not the application so if the desktop file is originally meant to define things at a per-application level it probably is not automatically the right place to define window properties.
Anyway I shall watch to see how the outcome of this affects applications like virt-manager (which I do use) since that application as part of its normal behaviour controls multiple windows representing:
- a list of guest machines accessible by a libvirt connection
-- a window containing the guest machine settings
-- a window containing the virtual guest display (whether graphical or text console)
(these two may share the same window via a toggle)
(there may be 0, 1 or multiple guest windows open at a time)
- a list of virtual networks with a dialog for editing a network's settings
- a list of storage pools each containing disk images. with dialogs for editing settings of a pool or an image
I'll need to check if it currently displays a customisable icon for each guest machine that has guest display open, since an app like virt-manager (or similarly virt-viewer) certainly does not know in advance at installation time what guests the operator will choose to deploy at runtime.
Don't give GNOME any ideas. They'll make it so you have to specify allowed window titles in your .desktop file or else the title is "Wayland Application".
@@thewhitefalcon8539no, "Insecure Application"
"Everyone will be happy". Lol, bruh, have you even seen 'everyone'? We have a whole site (twitter) dedicated to arguing over which things to argue about. If you've got two people in a room together, at least one won't be happy.
I can't believe I watched this whole thing 😅. What a crazy conversation. Thanks for another great video. 🙂
At this point Wayland is just a debate club that comes with a display protocol
Well, I have a desktop entry file for Firefox Developer Edition which I also use to start it, but when I start it, it switches the taskbar icon from the blue Firefox logo to the yellow W and when Firefox updates first it doesn't even stick to its taskbar icon, instead creating a seperate icon in KDE's task bar with the W logo on it. So if Firefox could at least set its own icon then it would be much nicer.
These are times where you see the value of a ship captain that has the authority to decide and tell : "It's gonna be this way, deal with it"
an application window just supplies its own icon? the end? wtf man. this is like 1990s OS stuff.
This reminds of c++ proposals. Sometimes new features are stuck in discussions for years, even if they have one or more implementations already provided, user feedback, bugs and quirks mostly resolved.
30 min video of a thread on github, you are a genius, love your work!
Btw I watched the whole video.
You should have covered the even-more-convoluted discussions about Bcachefs trying to get added to the kernel...
Excellent video 👍 Thank you 💜
Gimp uses this feature. The currently active image is shown in the title bar. Similarly it would be nice to have the favicon of a website showing in the titlebar.
God speed to Matthias (it really does feel like they're going up against the MCP from Tron quite often...), because this is absolutely insufferable... I had to double check to make sure this video didn't just start out as window icons but then branch into this other (actually) major thing, but nope... Just window icons. It's *window icons* for crying out loud - why does this need to be so complicated?
They bring up concerns of theming and unification, etc... but you know what tends to look really out of place on my taskbar? The generic Wayland icon. Hell, right now I don't know how, but Alacritty has no icon whatsoever, not even the generic Wayland one. The icon has just gone missing *everywhere* somehow.
I've seen it mentioned that the only way progress gets made on the Wayland side of things is if people use it and bring in feedback about what needs changing, but personally I wouldn't dare touch these discussions with a 90 foot pole. I could understand if this were a major thing, lets say if we were talking about compositors auto-updating apps for whatever reason - that would deserve a LOT of discussion to get it right. But this? Really?
Makes me glad I don't work in the application development space, I primarily work on backend stuff and occasionally mobile. Those worlds have some bureaucracy for sure, but this is a whole other level.
Finally, that remark about "I will make sure this gets a NACK in the end" comment is just... well I have a bunch of words that I'm going to keep to myself. I will however say that this sort of remark is why whenever people say that the GNOME devs are... lets just go with "something else", they get an "ACK" from me. Obviously, not all of them (and I respect quite a few of them), but as the saying goes it only takes a bad apple to ruin the bunch. Don't they have a code of conduct anyways?
With each passing discussion about _any_ sensible Wayland protocol I lose bit by bit any respect I had for gnome developers. I'm sorry, but it is always some random gnome maintainer who comes out of nowhere with what seems to be their life goal to drag and stall any meaningful development across the linux ecosystem.
TBF there was also a reservations from the KDE side that this wouldn't work for them at all either.
I wonder how did you get such respect for them, GNOME devs were always like that.
@@jfolz The comment from the KDE side was technical, about having to pass icons through several subsystems. The comment from the GNOME side was just ideological insanity, saying everything should be a .desktop file because he just feels that everything should be a .desktop file because it just should.
@@artemsmushkov766 Every since GNOME 3. By the way systemd and pulseaudio were also pushed by GNOME3.
@@thewhitefalcon8539 there's quite lot more to the discussion with Xaver Hugl that wasn't shown. The problem isn't "GNOME developers", but the recurring character Sebastian Wick.
Last time when they nacked something before even reading, we knew the whole Wayland thing is going to be its own issue in development... Cool new display protocol, great performance, not so great people there
loved the way you went over their drama while being respectful about it
Who needs DPI font scaling when you can have this?
just lost everything at the bringing up of CSD out of nowhere
I can't help but wonder how a PR by Miku-chan would go on Wayland's repos...
I'm just thankful being a user instead of being in development and getting caught in weeds of small feature developments. Designing features by committee is incredibly problematic, and means your feature, developed over a few days, will take 3 years of arguments before it can ship.
The first chip looking thing is just cosmetic without much meaning. The second chip looking thing shows the chip laid on a PCB and the third is the chip symbol as it would appear on a schematics. PCB design is split in two different stages the first is called schematics design and the second is PCB layout. The first deals with which two pins are connected together. The second deals with the physical placement of the chip. The first window would be the overall project management interface, the second would be layout editor and the third would be schematics editor.
This Wayland stuff is really challenging my open source enjoyment.
Wayland seems like a typical case of "design by committee"
EDIT: haha you said exactly the same thing at the end of the video!
Re: 24:20, that was partly my fault. One of the surprisingly vehement arguments against custom window icons was that we need forced .desktop icons to protect against phishing... which doesn't make sense when, for a transient window like a password prompt, the only place the user is likely to look that might contain an icon is the titlebar.
(I don't know about you, but I'm fairly security conscious and I certainly don't check my taskbar every time I get a password prompt.)
How funny, the simple "everything is a protocol" solution turns out to be more complex than what was a single call in X11.
Everything is a protocol in X11
I think I saw this writing on the wall before I switched to Windows after 15 years of trying to make desktop Linux work for me. GNOME and its developers basically aren't in the same headspace as the other desktop developers. Wayland needs to be forked for everything that isn't GNOME. Moreover I think forking of many freedesktop stuff also is needed because to make it work with GNOME many developers had to remove features that would otherwise improve the user experience on the Linux side. However there are two big problems on that road: Firefox and Chromium. Both browsers use GTK stuff to draw Linux-desktop interacting UIs and have access to things like printers. With Chromium the problem is even bigger since many apps are Electron, all of them need to be recompiled. Forking them and/or rewriting their UI and API calls is a huge undertaking which is what enables GNOME to dominate over the other desktop environments.
I agree, but why Windows!?
I am all for replacing X11 with Wayland, something I have been waiting to do for my own machines for 10+ years, and I think I will wait a little more.
as for the use cases, eom (an image viewer) makes an icon from the currently viewed image
I don't see any reason to use Wayland right now. To me, it seems like it doesn't have anything that X11 doesn't, and it's missing features that X11 has. Every window opens in the center of the screen, for example, because they don't have a way to put themselves where they were when they were closed.
21:06, paused here for a moment, why not just make the "single" icon name (which I presume is just a string for DMs to read, not for wayland itself to use) a combination of default_icon:callback_name where the callback_name refers to whatever callback is supposed to return the name of the icon to use for a given size, no lists, no altering wayland itself, just a standard that can fallback to just the icon name for apps that don't provide a callback. Incidentally I expect the callback itself to just look like this:
size_t callback( char8_t *dst, size_t max, size_t w, size_t h, size_t d );
If dst is NULL then the returned value should be the length of the filename that will be given, otherwise the number of characters copied.
**Edit:** btw I don't have an account on that site, tried making one but haven't received the email I'm supposed to get. So if someone wants to pass this suggestion onto the site, be my guest, go ahead and copy the YT link to it too if you think you'll mis-represent the suggestion.
**Edit 2:** some of you might be wandering about that size_t d parameter, that's for depth, gives the option of 3d icons
Should i implement this feature into my display manager before Wayland does? 😂
Yes
@@thewhitefalcon8539 After implementing the other stuff and fixing up the performance then.
Single thread software rendering with PHP goes quite slow after all...
@@blinking_dodo What kind of abomination are you trying to create?
@@michawhite7613 If Brodie doesn't forget, you will eventually see the video... 🙃
designed by committee ....
Poettering was right to push his vision through .
Imagine to discuss weeks and months over window icons wtf ...
Amen to that! The problem also is that core members of this committee don't want any changes. They want to keep the protocol as clean and pure as possible, in mint unboxed condition so they can admire how architecturally pure it is for 11 hours a day. Other peoples problems are not their problems. Other peoples solutions are.
I get the feeling that there's also a certain "we don't have/want this feature in *our* compositor, so no one else should have it either" mentality going on.
@@qwesx Wayland needs a guy that is willing to be hated by everyone and pushes the through what he thinks is needed. The problem is the commttee thing. A a guy volounteering as asshole for the purists and orthodoxy , with some knowledge on the matter and a vision would realy help everyone.
You know... I thought Wayland was finally going somewhere. With every video you do on Wayland, I become less and less sure of that. I some of the people involved in Wayland are like freaking teenagers in high school. Shameful for grown, intelligent (I'd hope) adults.
It's getting somewhere we just first need to reinvent legs
@@BrodieRobertson ...only for GNOME to veto it because they've been walking on their hands for years and can't see a valid use-case for legs.
When I left the Linux desktop community nearly a decade ago, dynamic window icons were fully supported in X11. It strains credulity that adding support for such a thing in X11's successor is somehow controversial, or even that it's not already solved. This is honestly sad.
I love the Wayland drama videos, Mr. Wayland
To be fair, at 12:40, my intent was just to point out that, if that route WERE chosen, a lot of the existing machinery could be reused and there was even precedent for it. (To try to eliminate "too much work for too little return" as a counterargument.)
When I get into these threads, I try to keep myself from getting attached to a specific solution and to focus on ensuring that downstream needs, pros, and/or cons are acknowledged. (eg. "Application A likely will ignore the solution you write for them if you don't provide capability B" or "If you do C, you can reuse existing part D".)
I will admit I should probably have made it more clear that, when I wrote "and have the compositor save the .desktop file it receives into it", I was imagining some kind of portal-like permission prompt being involved.
I'm glad all I have been doing the last few months is bug fixes on open source repos. OMG I can't imagine dealing with that. Funny, I had like 3 lines of code for a commit to fix an annoying bug and someone else who I honestly believe had a better solution that was both more elegant and directly fixed the issue got smacked down and my idea was implemented anyways. I should mention my method worked but just used if statements to exclude calling a function in certain scenarios. The other person solution directly fixed the issue. For whatever reason they decided a 3 line fix with no breaking changes and no API changes was better than adding multiple classes and possibly breaking other stuff. I honestly am glad I am not making those decisions.
I love Brodie slop
So Wayland is still incredibly half-baked after how many years of them telling us "it's ready to use"?
i would love to see everyone to come up with a solution for this use case... i see nothing wrong with application use multiple windows and able to set icon for the its window. kudos for Matthias Klumpp.
I think wayland has too many chefs in the kitchen. Sure the basics get done, but the fine details get lost in the chaos.
"May I suggest something that would better the user and developer experience" "Oh is this another one of those /controversial/ protocols!"
Which one is harder, contributing to Gnome or contributing to Wayland
Try both, take your choice xD
It's the same thing as long as GNOME has veto rights in Wayland.
I hope this isn't like off topic for the comments. I would love if you made a video about how perl became so widely used on linux. I see so many perl scripts on linux related projects and in distros. I would love to know more about why perl became such a big language for linux scripting.
The short answer is that it's a much better development environment for things like shell scripts, it has very good integration with UNIX tools and fits with the OS semantics and thus it excels at pipes, file slurping, inter-process communication, and other geeky tasks, its main strength is in text processing and it does that in a less awkward way than let's say sed or awk and CPAN is absolutely massive so you have everything that you could possibly want, maybe even multiple packages
I'm more than happy to take suggestions
I wonder if at some point a bunch of compositor implementers, like KDE or GNOME, will just fork the protocols project out of frustration and tell these bickering idiots to get bent. It wouldn't be the first time something like this happened. The W3C also lost large parts of their responsibilities to WHATWG because they where too slow.
GNOME is the one that's causing all these problems. KDE could do it, but KDE seems content to stick to the standards and only extend them at their stupid points, rather than making whole new standards like GNOME does.
GNOME is specifically the one which has held Wayland back all these years. KDE has been doing good work, but they've been too focused on working around the problem that is GNOME but playing within their rules, rather than creating a unified front with all the other Wayland developers who actually want to get $#!+ done. If they all had just kicked GNOME to the curb, Wayland would be finished already.
I'm assuming the three icons are Main Application/Library Manager, PCB Layout, and, Schematic Diagram.
25:14 Actually, isn't Enlightenment also CSD-only in its Wayland?
What I am thinking is that wayland should have a way for apps to pass info to the DE about the window to be created. Wayland should not care if that is about icon, color or whatever, that should be up to the app and the DE to do whatever they want. I think this solves the issue and whatever similar issue may occur in the future without the need for changing the protocol again. And the data structure should probably be in .desktop format just for not reinventing the wheel.
why can't a desktop entry file and any icon data, etc be written to the proc virtual filesystem and be removed from memory upon the application being killed?
Whenever I see controversies like this I'm convinced some Wayland devs are either on the spectrum or really enjoy their glass barbies. The idea of requiring .desktop files just to show icons for windows is absolutely mind boggling. I really hope X11 stays supported for years to come.
Sebastian is insinuating that he'd convince someone else to give the NACK. He's insinuating that he has "power" and "connections" to make it happen?
Could Wayland please stop trying to take on security responsibilities that are none of its business?
Wayland - It's not your job to protect me from apps plagiarising each others icons, just paint the damn picture on the screen and let me get on with my day!
They're probably still traumatized with x11
@@evildragon1774 : By this point, I'm certain they're traumatized by Wayland.
The concerns are somewhat warrented and considering how annoying per-window decorating is in Xorg its _slightly_ understandable its so controversial. However, what stops xdg-desktop-icon from supporting per window icon entries? What stops app stores from requesting all relevant icons before shipping? Why would it not be a good idea to let windows override their own window icons? And can we _please_ have unique sub-window names so we can theme and manage window icons at the window manager level? because that was attrocious on Xorg
If more users are on board wayland, the more we can steer it away from freebsd propaganda such as this