XDG Top-Level Drag Protocol Approved For Wayland

[Daktyl198]

So Wayland needs a protocol for every poop? Does not sound convenient nor sustainable for UI developement.

So did X11, you just never knew it existed. The entire reason Wayland was necessary is because X11 is a specification of a protocol, and X.org is an implementation of that specification. And they can't change the underlying code of X.org without breaking the specification.

The main difference is that X11 had a server that ran as a daemon, and the compositor was the client. Now the compositor is the server, and windows are the clients. This represents an interesting shift in how things work where a lot of things we used to do with the X11 spec don't translate 1:1 with Wayland, so new protocols have to be proposed and implemented rather than simply copied over.

[Myownfriend]

Now, if only we could get xdg-toplevel-icon approved...

I was racking my brain about that because I get why it's as controversial as it is.

Desktop files already provide icons for an application, which are theme-able, but they don't work for portable executables, applications that dynamically load plugins which have their own windows and icons, and doesn't necessarily work for Windows apps . They don't just provide icons though, they also provide other useful information about the application like it's Category, a localized application name and description, variables to launch the application with, and other stuff and it does this in a way that is display server protocol agnostic.

At the same time, if the applications provides all of their own icons, then the application might show a different icon in it's title bar (or in the overview) than the user clicked in it's launcher. Also windows with different icons than their application, might look like they're unrelated and the compositor wouldn't necessarily have the ability to communicate to the user which windows were spawned from what app. It would also only work with Wayland applications. Since its only a property, it kind of feels so divorced from the specifics of how the display server protocol works that it feels weird for it to be a part of it. At the same time it's window-specific so it feels like the best place for it to be.

Of course there also needs to be considerations for when the compositor needs an icon of a specific size or format, or whether it wants to use a symbolic icon.

[Myownfriend]

Hi, working through writing Wayland clients, active reader of fd.o stuff. Other answers are right enough, this MR will unify the DnD techniques with a best approach. ...since stopping fragmentation in compositors is a goalpost for developers using Wayland, I have to mention that the process in the story unifies them.

Thank you! I'm glad there were two people here to explain. I know the old DnD protocols won't be dropped for compatibility reasons (also I thought they were part of the core protocol) but can this new one be used as replacement for the old ones in newer software? Or would there still be reasons to use the old ones?

Also...

I feel like this is public service for said libertarian neckbeard wizards, coming from a communist neckbeard wizard.

I love this sentence. I welcome the communist neckbeard wizards lol

[daemonburrito]

Thank you! I'm glad there were two people here to explain. I know the old DnD protocols won't be dropped for compatibility reasons (also I thought they were part of the core protocol) but can this new one be used as replacement for the old ones in newer software? Or would there still be reasons to use the old ones?

Also...



I love this sentence. I welcome the communist neckbeard wizards lol

Thank you!

This is way better than anything I could write, from an actual contributor, KDE's David Redondo: https://blog.david-redondo.de/kde/wa...evel-drag.html It has interesting background on the Qt/kwin status, and going forward in Plasma.

[rmfx]

I prefer they take 9 months to add the protocol functions properly with long term design, rather than rushing a crap standard that will need a modified extension in a few months after they realize it was not made the best way.

[bitterseeds]

Hi, working through writing Wayland clients, active reader of fd.o stuff. Other answers are right enough, this MR will unify the DnD techniques with a best approach.

I feel like this is public service for said libertarian neckbeard wizards, coming from a communist neckbeard wizard.

This old neckbeard appreciates the explanation.

[theriddick]

If anyone asks why Linux hasn't already became the best desktop OS vs Windows, OR why Wayland isn't ready. You can point them to things like this.

Linux Desktop/OS: COLOSSAL potential, at a excruciating snails pace compared to proprietary desktop OS

[patrick1946]

At the same time, if the applications provides all of their own icons, then the application might show a different icon in it's title bar (or in the overview) than the user clicked in it's launcher.

Are not client side decorations made for that?

[X_m7]

Glad to see a Wayland protocol actually going through with just practical discussions instead of "oh no it looks like X11 get it outta my face" stuff for once.

[X_m7]

Are not client side decorations made for that?

The title bars would be covered by CSDs, but not the icons in the overview, taskbar/panel, Alt+Tab app switcher, and wherever else where an app may show multiple entries because it has multiple windows that is outside of the app's windows.