Xfce 4.16 To Drop GTK2 Support, Explore Some Client-Side Decorations

[duby229]

Sounds reasonable. CSD proved to be the way forward.

That's bull. What CSD proved is it breaks look and feel, most devs refuse to duplicate common window management capability, it increases white space and reduces usabilty. None of which is a good thing.

[Britoid]

That's bull. What CSD proved is it breaks look and feel, most devs refuse to duplicate common window management capability, it increases white space and reduces usabilty. None of which is a good thing.

I dunno about you, but with running applications targeted at a different desktop, the different titlebar is the least of your problems.

Anyway, as previously said, headerbars != csd.

[duby229]

Too late, they already said yes.

Time to use something sane like LXQt or Plasma then.

Already switched to plasma long ago. Well worth it. It's by far the most common sense UI in existence right now.

[duby229]

I dunno about you, but with running applications targeted at a different desktop, the different titlebar is the least of your problems.

Anyway, as previously said, headerbars != csd.

Except that in all cases CSD is used the default (and many times only) option is to hide the titlebar. So yes it does 100% it does.

(EDIT) You do know what window management is for right?

[duby229]

CSD has no redeeming value. the inconsistent look and feel alone should make it a no brainer. The hidden titlebar by default should make it a nonoption for everyone. The increased whitespace and decreased usability should make programmers rethink it.

(EDIT) Also worth mentioning, basically all third party CSD apps I've ever used override the desktop theme anyway and that just exacerbates the inconsistent look and feel problem that CSD forces on you anyway.

[duby229]

duby229 Britoid already told you headerbars != csd. And that was also stated in the xfce blog.

Except that in All extent scenarios, yes it does...

i'll wait to see what xfce does, but precedence is what precedence is.

[Britoid]

Client side decoration is where the client application is somehow responsible for drawing the window controls. This is quite the norm on platforms like Windows, macOS, Android etc because you don't usually talk to their compositors that much directly, you talk with the toolkits/libraries (which handle the decoration) which themselves talk to the compositor

In X however, window managers were the norm, however we're not in an X world anymore. Imho on GNOME, Qt, SDL etc should talk through GTK to create windows (there's already patches to do this) in a similar fashion to above. However, I don't believe this is quite as efficient and abstracted as it could be, maybe dmabuf may help here but I'm not an expert on the graphics stack.

Headerbars are a design decision, this is evident because under GNOME Wayland, if a GTK app does not use a headerbar, it still has it's decoration draw by GDK, hence "client-side".

[Mavman]

Time to use something sane like LXQt or Plasma then.

Totally agree!

For a regular computer, Plasma without any doubts.
For old or special machines where all resources count I definitely prefer LXQt!

Both avoid CSD and I couldn't agree more with the choice!

[flower]

Totally agree!

For a regular computer, Plasma without any doubts.
For old or special machines where all resources count I definitely prefer LXQt!

Both avoid CSD and I couldn't agree more with the choice!

my only problem is kio. if plasma would integrate something like kio-fuse deep enough (eg i can copy file links and have the fuse link - not the kio one) i'd happily use it.
i dont know about lxqt though. do they use kio too?

[Venemo]

I didn't imagine people would be so passionate about who draws their title bars. Why does it matter so much, as long as they serve the same purpose?