Gräßlin: Client Side Window Decorations and Wayland [LWN.net]

KWin hacker Martin Gräßlin discusses client side decorations (CSD) and Wayland on his blog. He notes that while Weston—the reference Wayland compositor—requires CSD, nothing in the Wayland protocol does. "I had a talk with Andy from Qt Wayland fame about the CSD implementation and he explained [to] me that inside Qt the CSD code gives some overhead and that they have a flag to turn them off. Which is great. And we in KWin already have server side decorations and will need to keep them around for legacy X applications. What's the point then to use CSD in Qt if we already have the decorations and can give the application a better performance? Well none and that's why I plan to use server side decoration in KWin on Wayland."

to post comments

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 14:49 UTC (Fri) by cmorgan (guest, #71980) [Link] (83 responses)

I've been following Wayland for a while and this kind of thing has come up several times on the mailing lists. As one person who responded to Martin's post mentioned, what about consistency in the UI for the applications themselves? Why the concern about decorations?

Imo there is a bit too much concern about CSD. Toolkits might want to use a common library to ensure that decorations look the same but client side gives more flexibility afaics. Some applications might want unique looking decorations, developers can appreciate the ability to innovate in that area. If not they can just pick an appropriate toolkit that does what they want.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 15:15 UTC (Fri) by gnb (subscriber, #5132) [Link] (13 responses)

> Some applications might want unique looking decorations

And some users see that as a problem: I've set up the window decorations on my machine the way *I* like them (maximise and minimise all the way to the right, close all the way to the left a long way from everything else), why would I want a proliferation of applications ignoring that?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 15:27 UTC (Fri) by cmorgan (guest, #71980) [Link] (9 responses)

I quite agree. It is helpful, as an end-user, to push developers to support standard decorations/themeing etc. I just advocate that approach vs. trying to force it server side. The downsides of doing it server side seem severe, what would you do if you wanted to change decorations in the case where things were forced server side? Its added complexity to then have the ability to switch it off etc.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 17:32 UTC (Fri) by dlang (guest, #313) [Link] (8 responses)

the point is that if you want to change things and they are done server side, you have one place to go to make the change, the server.

If they are client side, then you have to reconfigure every app (or at least every toolkit that an app may use), assuming that the app/toolkit gives you that option, most apps won't bother

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 18:20 UTC (Fri) by cmorgan (guest, #71980) [Link] (7 responses)

That's the downside of having the graphics system not deal with decorations, everyone has to do it themselves. It may seem like anarchy but most people will, as you've mentioned, use one of the popular toolkits (Qt/gtk etc). Toolkits already work together on theming/colors, although maybe not as well as they probably should. Doing CSD doesn't mean that existing work to coordinate look and feel is discarded, it does however simplify the graphics system and let people who care about custom decorations implement them without as much effort.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 16:30 UTC (Sat) by ovitters (guest, #27950) [Link] (6 responses)

I'm not aware that Gtk+ works together with Qt regarding theming? Gtk+ can be made to look similar (as Qt), but that is not done by any sharing of themes. Best is to make a Gtk+ and a Qt theme, so duplication of work.

Correct me if I'm wrong (no theming expert at all)…

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 22:20 UTC (Sat) by set (guest, #4788) [Link]

I mostly use gtk+ apps with custom theme. I told qt to use the 'GTK+'
theme, and the qt apps look like the gtk+ ones.... so it looks at the
specific gtk+ theme being used and mimics or adopts it.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 10, 2013 13:27 UTC (Sun) by Company (guest, #57006) [Link] (4 responses)

We (GTK+) do not work together with anyone (Mozilla, Qt, Chrome, Enlightenment, you name it) on theming. All toolkits have their own more or less esoteric theming languages. And I don't think that's gonna change, theming is too integral for the toolkit to standardize (just compare GTK2 and GTK3 there...)

So all similar looks are usually the result of hard-working theme designers.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 10, 2013 13:57 UTC (Sun) by HelloWorld (guest, #56129) [Link]

Yeah, that must be why QGtkStyle works perfectly fine. To be honest, I just wish that Gtk went away and never came back. Having multiple GUI toolkits offers no benefits, makes user interfaces inconsistent and increases the maintenance cost.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 11, 2013 13:11 UTC (Mon) by nye (guest, #51576) [Link] (2 responses)

>We (GTK+) do not work together with anyone

That pretty much sums it up.

I too think the world would be a slightly better place if Gtk went away and never came back.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 12, 2013 1:02 UTC (Tue) by Company (guest, #57006) [Link] (1 responses)

Because E17 and KDE use a Freedesktop theming format I've probably never heard of?

I mean, are you just trolling for no reason or do you know any toolkit that shares anything theming related with anyone else?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 12, 2013 16:23 UTC (Tue) by nye (guest, #51576) [Link]

>I mean, are you just trolling for no reason

No, I think it is a statement of simple fact that GTK developers are less interested in compatibility than providers of other toolkits, or at least Qt, which is its principal competitor.

For example:
Qt has a toolkit style which uses the Gtk theme, in order to integrate as well as possible. It's not *absolutely perfect*, but in practice it works very well.

Amongst other things, Qt applications using this style will automatically use the native Gtk file dialog, so you often wouldn't even know that you're using a Qt application on your Gnome desktop.

There *is* a Gtk style to do the reverse, but this came from the Qt side, not the Gtk side, because the Qt side is interested in integration and compatibility, whereas the Gtk side is not.

The same idea extends to other platforms: on Windows, Qt applications are largely indistinguishable form native applications, including things like using the native file dialog. I'm led to believe that the same is true on OS X, though I've not personally tried it.

Gtk applications on the other hand use a style similar to, but distinguishable from, native widgets, and worse, they impose their own file dialog.

The problem goes further than simply theming.

Qt is able to use the glib main loop, in order to allow more options of component integration; Gtk to the best of my knowledge has nothing like this flexibility (I welcome corrections if this has changed).

Generalising a little further, I'm going to conflate Qt/KDE and Gtk/Gnome; disregard this part if you think that's unreasonable:
The Qt/KDE camp puts greater stock in interoperability; that doesn't just mean in providing applications that fit in better on other OSes as in the example above, but also in working with standards on Free desktops. In principle, freedesktop.org exists to further interoperability between Free desktops by defining agreed standards, In practice that has a tendency to mean 'Gnome is doing this; everyone else must too'. When a standard didn't originate on the Gnome side, situations like the StatusNotifier problem arise: http://aseigo.blogspot.co.uk/2011/03/collaborations-demis...

As a rule, Gtk and Gnome effect a 'my way or the highway' attitude, reasoning that because together they are the 800lb gorilla in the Free desktop world then interoperability consists of people bending over backwards to suit them.

It is my belief that Gtk's continued presence despite the existence of at least one freely licensed and technically superior alternative is ultimately harmful to the Free desktop, in part because of the inevitable fragmentation of interest/effort/etc, but mostly because Gtk eschews standardisation in favour of control, and places ego above technical excellence, ultimately resulting in applications which are - in a small but non-zero way - not as good as they could have been.

The existence of Gtk is essentially a historical accident; if Qt had been freely licensed from the beginning, then perhaps the GIMP would not have spawned its own toolkit. Even if it had, it's unlikely that an entire desktop environment project would have grown around it. Given that the Qt licence problems were solved a decade past, it's unfortunate that Gtk continues to exist.

(I keep mentioning native file dialogs because the Gtk file dialog is indefensibly awful. It's so bad that simply trying to use it actually makes me *angry*; I can't imagine how anyone could produce something so brain-damaged and not feel a deep sense of shame.)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 16:43 UTC (Fri) by drag (guest, #31333) [Link]

In practice I expect most application developers will just stick with the toolkit defaults, whatever they may be, unless they have good reason.

If application developers wanted funky decorations they could have that right now, in X. A few examples of this would be 'conky', 'xmms' (and dozens of other media plays), gkrellm, docky, gnome2 panel.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 5:34 UTC (Sat) by daniels (subscriber, #16193) [Link] (1 responses)

There's an easy solution there: don't run substandard apps.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 12:27 UTC (Thu) by robert_s (subscriber, #42402) [Link]

All apps I use are "substandard" in one form or another.

That doesn't mean I have the luxury of not using them and still getting useful work done.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 16:13 UTC (Fri) by error27 (guest, #8346) [Link]

That was addressed in the blog post. In theory lots of people could do stuff with window decorations, but in practice only Chrome does.

But the problem of doing ugly stuff with decorations is real already.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 11:27 UTC (Sat) by Seegras (guest, #20463) [Link] (7 responses)

> Some applications might want unique looking decorations

No. Some idiot developers might want to shove unique looking decorations up my arse. This is a completely user-unfriendly idea. If they want something "to look cool", they should release desktop-themes and widget-themes instead.

I've been there; I'm running Linux since 1995, where EVERY application used its own toolkit and widget-set and had a completely different look & feel (I had such a widget-horror screenshot somewhere, but I think I lost it).

I want to be able to set the widgets and decorations globally, whith no application even being able to override it (Note that this is quite different from Windows and MacOS X, where some company decides for me my that my desktop has to look like fisher-price or something).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 16:04 UTC (Sat) by mpr22 (subscriber, #60784) [Link]

Decorations? Sensible. Widgets? The opposite of possible, because one of the widgets that the Universal Widget Tyrant has to provide in order to get anyone at all ever to target it is a canvas modifiable on a per-pixel basis.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 9:43 UTC (Tue) by fdrs (subscriber, #85858) [Link] (5 responses)

And don't forget how awful Java Swing applications look _anywhere_ ...
I agree, window decorations _must_ be a platform defined (_and_ enforced).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 11:01 UTC (Tue) by mpr22 (subscriber, #60784) [Link] (3 responses)

I think that window decorations look terrible on application splash windows (and yes, splash windows do fulfil a useful purpose; they give the user confidence that their slow-starting application has in fact started to start, while avoiding the "here is your interface, even though your application is not yet usable" problem; I think every GUI application that takes more than, say, 500ms to start should have one).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 15:24 UTC (Tue) by paulj (subscriber, #341) [Link] (1 responses)

I really hate splash windows which insist on being in the foreground and that can not be dismissed, *especially* if the app takes a long time to load! I could be doing something else, if not for the splash window!

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 15:29 UTC (Tue) by mpr22 (subscriber, #60784) [Link]

Those are bad too, yes.

Gräßlin: Client Side Window Decorations and Wayland

Posted Mar 3, 2013 20:40 UTC (Sun) by JanC_ (guest, #34940) [Link]

Why not show a normal progress dialog instead of a "splash screen"?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 15:21 UTC (Tue) by nybble41 (subscriber, #55106) [Link]

You can't _enforce_ a particular style of window decorations unless you're willing to require all windows (including specialized windows, like pop-up menus, panels, docks, etc.) to have a border which fits inside the screen. No doubt someone will write a Wayland compositor with that sort of restriction, but I doubt very many people will run it.

If Qt and GTK use a common library to handle client-side window decorations, as planned, the result will probably be at least as consistent as the current scheme under X, with the added benefit that the application and the compositor don't have to coordinate drawing into a single buffer, or make an extra copy of the window contents at the end of every frame to add decorations.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 16:13 UTC (Sat) by hitmark (guest, #34609) [Link] (2 responses)

The curse of the media player?

Seems like all of it hankers back to trying to recreate winamp on *nix...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 16:33 UTC (Sat) by drag (guest, #31333) [Link] (1 responses)

Actually it's just a bunch of dedicated hackers who have spent decades of effort trying to improve X Windows and related systems only to realize the best approach is to try to do something different.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 18:54 UTC (Sat) by dtlin (subscriber, #36537) [Link]

Highly relevant, from this week's edition: http://lwn.net/Articles/536487/

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 13, 2013 22:52 UTC (Wed) by mmarq (guest, #2332) [Link] (56 responses)

There isn't only too much concern about CSD, there is too much concern about toolkits...

Many years of opendesktop.org... yet when do we see a *COMPLETE* Icon set theme that can be applied to all ?? ... and in here i mean Mozilla side and Libre/Open Office side ??

This would do much more for consistent look&fell than any toolkit... being the worst of all that most (for not saying all) of icon set themes are incomplete.. leading sometimes to bizarre looks.

Also a common color engine and a basic theming engine is all that suffices, no matter what "lib" renders them, as long it doesn't mess things. The *same* textures an or pixmaps could only end up looking very alike.

Today the best option is go *individually* trying to match all the different themes in a away they look the same... a nerve breaking journey to say the least.

The rest are tempests-in-tea-pots... let them have the away they like.

But if the rendering "side" and toolkits are a concern, what about to say about wayland ? ...wouldn't it make things quite more difficult... if Gnome... or Mozilla and Libre/Open Office stick with X... and the other half passes to wayland(besides a stab to dead to many network stuff) ?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 13, 2013 23:22 UTC (Wed) by dlang (guest, #313) [Link] (55 responses)

The reason people are opposing client side decorations is exactly because there are so many toolkits, and we don't believe that they will ever be consistant and use the same theme.

Currently, with the exception of the small number of programs that insist on CSD, this doesn't matter because there is only one place that needs tob e configured to manage window decorations, in the window manager (yes, different window managers require going different places to make changes). All windows (almost) created on a display will have the decorations in the same places.

With CSD being the norm, the expectation is that different toolkits will choose to do different things and users will have to figure out how to close each application's window differently, depending on what toolkit was used

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 13, 2013 23:54 UTC (Wed) by khim (subscriber, #9252) [Link] (10 responses)

With CSD being the norm, the expectation is that different toolkits will choose to do different things and users will have to figure out how to close each application's window differently, depending on what toolkit was used

And this is such a big problem because 99% of time people are spending doing nothing but moving windows around and closing these pesky windows.

Seriously: applications handle in-window UI themselves and do it in somewhat inconsistent way - and this is not a problem apparently (there are still some users left, right?). Why is it such a big problem with windows decorations?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:37 UTC (Thu) by dlang (guest, #313) [Link] (4 responses)

> And this is such a big problem because 99% of time people are spending doing nothing but moving windows around and closing these pesky windows.

no, this is such a big problem because the impact of hitting 'close immediately with no second chance' on a window when you mean to hit 'maximize' is so drastic.

High impact problems that happen less frequently are still major problems.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 8:09 UTC (Thu) by khim (subscriber, #9252) [Link] (3 responses)

this is such a big problem because the impact of hitting 'close immediately with no second chance' on a window when you mean to hit 'maximize' is so drastic.

And the same problem always existed which keyboard - yet somehow people are not rioting and are not writing petitions to make Ctrl-Q illegal.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 10:45 UTC (Thu) by hholzgra (subscriber, #11737) [Link] (1 responses)

Exactly one of the reasons why i stopped using my Mac ... Cmd-Q (when using the right side Cmd key) is just too similar to AltGr-Q on a German PC keyboard ... quite a few long email replies have vanished as right before sending i figured out "i should put somebody@example.org on Cc:" and hit the wrong @ combination, closing the reply window without any confirmation dialog that way ...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 13:53 UTC (Thu) by Cyberax (✭ supporter ✭, #52523) [Link]

I remapped all functional keys on my Mac :) To confuse the potential adversary and to make it a little bit easier to actually press commonly used combinations.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 22:56 UTC (Thu) by apoelstra (subscriber, #75205) [Link]

> And the same problem always existed which keyboard - yet somehow people are not rioting and are not writing petitions to make Ctrl-Q illegal.

On dvorak, Win+J (switch to lower window in wmii) and Ctrl+Q are immediately beside each other, and feel identical on my keyboard. So I routinely close programs that I was trying to switch away from.

The reason that I'm not rioting or writing petitions is that it's my own fault, not that it's not a problem.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:44 UTC (Thu) by mmarq (guest, #2332) [Link] (4 responses)

ummm... enforce strict rules which would be *desktop environment very specific*, doesn't matter if CSD or SSD... and no discussion about it, and bickering about toolkits or semantics needed.period

Don't like the default ? ... there are some themes(could be plenty), else do your own or ... off.

See !?... i think this part is so easy...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 8:11 UTC (Thu) by khim (subscriber, #9252) [Link] (3 responses)

Any plan which includes word "eforce" is doomed to fail in FOSS word. Simply because there are noone who can enforce anything as was already said.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 14:36 UTC (Fri) by mmarq (guest, #2332) [Link] (2 responses)

Depends on what you mean by "eforce"...

It could be that is coded in a certain way that only permits certain possibilities, namely windows decorations being rendered in a certain way buy certain rules.

Who controls the "renderer" controls the look, unless with CSD you mean allow any kind of renderer... and then complain why certain apps apply their own and its a "looking" mess -> crazy.

so its not like "gun point" and you have to abide, but simply coded to use what is included and no other option.

Don't like it ?... you can always fork KDE or Gnome (etc)... or pertinent parts and do it your way, only don't expect others to maintain it.

what is wrong or difficult with that ?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 16:01 UTC (Fri) by khim (subscriber, #9252) [Link] (1 responses)

what is wrong or difficult with that ?

Well…

KDE or Gnome (etc)

That. There are always other options. And bikeshedding law means that the least important and complicated pieces will always differ.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 16:18 UTC (Fri) by mmarq (guest, #2332) [Link]

There is nothing wrong... you are talking in circles (semantics)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:25 UTC (Thu) by mmarq (guest, #2332) [Link] (43 responses)

Well in my limited knowledge of the matters

Toolkits evolve with new versions, right ?

Wouldn't much more easy and much more peaceful, to agree to have a common color engine and a common *basic* theme engine (-> could be designed to be extensible or embeddable in the theming engines that the toolkits might present), than to force everybody to use the same toolkit or the same toolkit overall guidelines ?

It wont provide *exactly exactly* the same look&feel... i think... because they will have different rendering engines, but it will be pretty close, miles better than now... i think...

I think the end result is not perfection (can't be)... or the perfection of the code, or for the developers much concern, because the end result is exactly for the ***users***... the idea;

* apply *ONE 'basic' theme for all* (at choice), which could be even set before entering the desktop, at the DM login phase (GDM KDM)
(the rest could be very desktop version specific, meaning wont touch panels special widgets etc))

* the same with Icon Sets

* windows decorations should be free, as is today... stupid to say the least to enforce anything here... i think... unless you don't supply any theming at all with new window manager.

Why *can't* it be done, **INCLUDING** Mozilla and Libre/Open Office ??? ... and why not JavaFX for all those GUIs based on its scripting languages, meaning target Android environment also ??
(for crying out loud opendesktop.org is a decade old, yet the only i can see is bickering about semantics and tastes on KDE vs Gnome)

< most of that is applicable but by hand and piece meal with the help of individual theme packages>

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:30 UTC (Thu) by dlang (guest, #313) [Link] (42 responses)

it can't be done because you can't get everyone to agree on what the one common color combination should be.

You are also never going to get every toolkit out there to use the same definition for where to look for information and icons, etc.

This is especially true because to agree on things, some toolkit is going to have to change what it does, which is probably going to break existing setups.

This isn't Microsoft Windows where everyone just accepts what is provided to them from the one true source. That is both a weakness and a strength.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:48 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link] (2 responses)

Toolkits can definitely change where they look for resources if there is agreement on a standard location. Look for the standard location first and fallback to using the older location with warnings for compatibility. It is not a unsurmountable problem except for petty politics.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 2:41 UTC (Thu) by mmarq (guest, #2332) [Link] (1 responses)

Please don't take me wrong... but just imagine...

Google invites the leaders of KDE and Gnome... and some others, including E...

"now gentleman you don't leave this room until you agree on some basic theming including android... pum!... bag of money on the table"

As soon as the host leaves the room starts a terrible battle... or how long it would take to agree on a lot of things ?? (lolz)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 19, 2013 9:51 UTC (Tue) by fdrs (subscriber, #85858) [Link]

I guess more time then Google would take to make one himself (Hi android...)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 1:52 UTC (Thu) by mmarq (guest, #2332) [Link] (5 responses)

"" it can't be done because you can't get everyone to agree on what the one common color combination should be. ""

lol... that bad ? ... for something so... little important...

Just imagine!... can't agree on something so trivial... color combination! lol ... more so cause end users end up changing to the colors they like... i just imagine trying to agree on something less trivial lol...

Thank you for the warning... if by any tumble of chance i happen to get invited to a meeting of opendesktop.org, can't forget to take a gun... lol...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 8:00 UTC (Thu) by khim (subscriber, #9252) [Link] (4 responses)

Just imagine!... can't agree on something so trivial... color combination! lol ... more so cause end users end up changing to the colors they like... i just imagine trying to agree on something less trivial lol...

Actually it's easier to agree on something non-trivial. Much, much, MUCH easier. You see, to implement something non-trivial (let alone complex) you need someone capable. If it's something not just non-trivial, but outright complex then this means there are very, very few opinions because there are very, very few people! And this means they can meaningfully discuss things and agree to do something. But everyone can change the color...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 14:56 UTC (Fri) by mmarq (guest, #2332) [Link] (3 responses)

damn semantics!...

Actually is extremely simple. You *think* about the end user... ALL OF THEM... not only your "followers" and or what you think is the best or the right thing...

This right mind set and it will be incredible easier to agree in a lot of things, its **for the "user"** is not for my particular project or way.

If are in the business to be praise by that "cool" comment of the closest followers, you are not in the business for mass adoption you are in a "RELIGION" business.

And those parallel examples are striking because only in the "christian" side as example, there are more than 2000 denominations... any of them claiming to possess the light and all others will end up in hell...

And then you complain about fragmentation and when it will be the year of Linux desktop... and seek for more support for your cause -> CRAZY!...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 15:23 UTC (Fri) by mpr22 (subscriber, #60784) [Link] (2 responses)

Spending more than five seconds thinking about the end user usually leads me to the following conclusions:

Some of them (I'm tempted to say most, but...) want "exactly what I have, only better" - but can't be induced to give you any information about what might constitute "betterness".
Some of them know what they want, and it's something reasonable and plausible that can be achieved on a sensible timescale.
Some of them know what they want, and it's either self-contradictory or in violation of the laws of physics as we know them.
Some of them want things that cannot be reconciled with things that other people want.
Some of them have no idea what they want.
Some of them know what they want, but will tell you they want something completely different that only resembles what they want in the way that Bizarro resembles Superman.
Some of them think they know what they want, but will hate it when you give it to them.

(I wouldn't be surprised to find that I have myself fallen into each of the above categories at least once.)

If we look at the platforms that have achieved mass adoption, we find that they've all said "no" to at least some physically-possible ideas. This suggests a guiding principle that being able to say "no" is important.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 16:34 UTC (Fri) by mmarq (guest, #2332) [Link]

7 billions and no face is exactly the same... damn sure the Large majority doesn't want what you want...

But yes there are common denominators, but those tend to be very small and restricted...

Users want some more (or more) consistent look & fell... "I THINK SHOULD BE ALLOWED OR BETTER SAID FACILITATED", and as i POSTED ABOVE, a minimal set of issues;

hey! KDE, hey! Gnome, hey! Mozilla, hey! Libreoffice... (etc) its not for notoriety is not to discuss better code, its not for my project gain... "ITS FOR THEM"... THINK NOT PERFECTION... BUT FOR SURE THERE ARE SOME BASIC THINGS WE COULD AGREE FOR THE BETTERMENT OF THEM AND US

That is a "Free Leader"... else, "free" is an empty word in the mouth of a fundamentalist sectarian kind of priesthood.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 18:48 UTC (Fri) by hummassa (guest, #307) [Link]

As a person who had to take user-centered design classes and works with it, I can tell you that: most users know what they want, but they can't communicate what it is. You have a plethora of communication techniques just to extract this information from a group of users.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 2:08 UTC (Thu) by mmarq (guest, #2332) [Link]

"" This is especially true because to agree on things, some toolkit is going to have to change what it does, which is probably going to break existing setups. ""

lol.. setups that the "end users" end up breaking anyway with every new release of an environment, importing themes of every kind right and left, sometimes quite unstable, and applying configs that none developer thought off or tested ( hey! why doesn't this or that appear!?.. lol)...

for a common *basic* theming it would only happen once, then everybody should be free to evolve around that as they like... and then business as usual, every new release and the user changes completely the look from the default lol... boy! that must piss terribly the developers lol</sarcasm>...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 3:18 UTC (Thu) by mmarq (guest, #2332) [Link] (31 responses)

oh! forgot ... setups that the developers end up breaking themselves with almost every new release...

I registered here more than 10 years ago... going trough all the pertinent forum archives i would only need 1 buck for every complain i find of ""my older setup this and that appear here or there and it did this and that and doesn't do anymore"" to be a millionaire...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 17:39 UTC (Thu) by nix (subscriber, #2304) [Link] (30 responses)

With the single exception of the abomination which is XMMS, and Chrome (in which you can at least turn this behaviour off), my window decorations and wm shortcut keys have been completely consistent and almost unchanging since 1994, and I use them all the time. These things are wired into my brain. Break them, you break a two-decade-old working pattern.

Needless to say, I will never use a system which enforces client-side decorations. *Never*. No matter what its advantages, the disadvantages to my workflow would be catastrophic. A bit of theoretical flicker that I never saw even on the slowest machines I used in that period is just not worth it.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 16:15 UTC (Fri) by mmarq (guest, #2332) [Link] (29 responses)

CSD or SSD is orthogonal, who controls the code that leads to rendering controls the "look", and there could be only one renderer/compositor for both CSD or SSD, which in X is already "detached".

Of course many could hack around this, and allow any number of compositor/renderers for that "Publicity Prone CSD Idea"... as many could hack around almost anything... but why should any "desktop project" bother a tiny bit about it ? (let them do it, only don't maintain it, they should)

And this goes both sides, changes are you adopted a piece of tech that as example allows any number of "renderers" hardwired in code (wayland)... hey! but you can also hack around(or off) this, if its OUTSIDE of what is "FOR ALL OF THE END USERS CONCERN" and "SHOULD BE ALLOWED"... "YOU DON'T FALL FOR MANTRAS" and damn sure hack them in their face(for not saying fork)...

Yes is important to maintain a "familiarity" legacy concern, that is IMO the MAJOR strength of Windoze, despite too much permissibility in certain areas for many tastes.

About the Wayland way, i think is not of concern for now, if DEVS ARE NOT STUPID... the code is PRE-ALPHA STATE... when it will be REAL, i mean, *REAL* FULL feature set, and *REAL* production stable ready... it will take at least a couple of years more... and this without counting any "network" features, because if you count them (as is circulating a possibility, is not specified what shape format or protocol(s), only the possibility), then you can count at least the double (4 years).

I thing Collabora suspects this, and also its main devs, only i suspect they think they can accelerate the process tremendously... that is why the SUPER OVERHYPE, and HATRED FOR X that transpires( when there are more X systems than xorg).

But has with many projects in the past... they(wayland) have a couple of very good features "ON PAPER"... this things much much more usual than not tend to NOT live to full expectations and deliver the results everyone/most 'hype' expected.

I trace a parallel with "Reiser", very good FS indeed, but when the dust settled is now bench provable its not the best and or the fastest FS around(not even second)...

The same will happen to wayland... i'm affraid...

The only thing really disgustable, is that they don't seem to "ISSUE WARNINGS ABOUT PRODUCTION READINESS"... some comments of users already circulates of "can hardly wait for this"... this suckers end up installing it on their production systems, and crash and burn horribly, and then blame everybody except their stupid gullibility and the *buggy* they installed thinking it was "so cool"...

This is the kind of thing that helps nobody, not even Collabora, because those kind of "suckers", rarely if ever helps in debugging, more so because they tend to not have the minimal knowledge and or skills required even if they intent to do it(which is almost never, only complain)... disgusting...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 18:11 UTC (Fri) by nix (subscriber, #2304) [Link] (21 responses)

HATRED FOR X that transpires( when there are more X systems than xorg)

Er, yeah. A good few of the Wayland developers were working on X even back then. I suspect this collapses your argument.

They hate the X server codebase because the X server codebase is hateful, not because they are irrational lunatics (except insofar as working on the X server codebase for so long has converted them into irrational lunatics, that is!)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 18, 2013 22:55 UTC (Mon) by mmarq (guest, #2332) [Link] (20 responses)

You are missing the point

And every other X system around?... is the code base also "hateful"(many engines were and a couple still is proprietary.. i think..) ??... have Wayland developers worked on them all ??

It might be that the NICE "BUT OVERHYPED" features that Wayland intends to implement, where already implemented elsewhere... and if all of those X engines were still commercial and going well, wayland might find itself in software patent troubles...

hey!... perhaps that is *the* reason they want to finish it off... but i tend to believe is to catch the most installation base the fast as possible, irrespective of who might crash and burn miserably... DISGUSTING

http://wayland.freedesktop.org/faq.html ... DISGUSTING of biased and "untrue"... sounds like HATRED...

This serves to show that there isn't only the Wayland way for possible similar features... the thing would be solved by "facilitating" the end user install the windowing system they like... it could be that the "same" window manager (like Kwin) could work with more than 1 windowing system...

If wayland will be such upper duper it will win without much effort... if not, just history repeating itself...

Forcing by "PROPAGANDA" is a way of windoze world... traditionally Linux has been about choice.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 18, 2013 23:41 UTC (Mon) by anselm (subscriber, #2796) [Link] (17 responses)

If wayland will be such upper duper it will win without much effort... if not, just history repeating itself...

It seems that the issue will sort itself out automatically over time since various important X developers also think Wayland is a good idea in principle. So far nobody is pushing Wayland down anyone's throat; chances are that Wayland winning out »without much effort« is exactly what is going to happen eventually.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 1:59 UTC (Wed) by mmarq (guest, #2332) [Link] (16 responses)

Eh!!!... and how do you change ~ 200.000 applications from any ties to X to be "tied" with Wayland ???...

As i said, X without network is not X ...

Wayland with X is just "another" X ...

Wayland is absolutely a gargantuan enterprise... to change an enormous app base to its side...

And that is anther disgusting aspect... they don't seem to care at all... its about at least catch a good desktop environment and put everybody in there targeting Wayland... ITS NICE... and fits the ideas of Android of online AppStore... that is the motive of Wayland and why they choose to be a clear departure... a break away -> PUT EVERYONE CODING TO ULTRA-MOBILE...

But i don't know how much more **MIOPIC** that vision can be... there is already X for mobile... and even this WILL BE NOT NEEDED...

4G networks... the norm now can go 100Mbit/s... ARM(the leader) has already launched ARMv8 64bit... in 2 years all superphones(and so tablets even better) will be 64bit, with FullDH screens, which some already present... ARMv57 is 5X more performant than ARMv15 by reference design(and more power efficient), some variants like Qualcomm Snapdragon or Samsung Exynos can push that even further...

Conclusion even a superphone in 2 years will have the same processing power of actual many even top AMD or Intel i7 desktop offerings... and GPGPUs capable of of considerable compute power, and more powerful than the actual current PCIe adapters used in mainstream( not talking the top lines here)... more so because AMD/ATI is also in the ARM boat(-> they have a license to ALL ARMv8 64 bit including big.LITTLE)... nVidia also, so the battle will resume...

The only problem is that the RESOLUTION can be the same, but the "SCREENS" tend to be much smaller... but in here i think Plasma Active could do a good job adapt... and then you can run an even "heavier" KDE (or Gnome) than you do now on the current desktops, without any problems at all (AT ALL)(more so flash 128Gbit is on the corner, 2 of those and a Superphone could have a "disk" with 64GB (yes GB) of space-> how much is FULL KDE ??)...

Wayland its better!?... ok i'm always for better, since it doesn't break existent stuff... but FOR MOBILE YOU DON'T NEED WAYLAND AT ALL... CONCERNS MIGHT BE FOR CURRENT MOBILE GPUS AND ARMv9... BUT THOSE WILL BE *UBER COMPLETELY OBSOLETED* MUCH FASTER THAN WAYLAND WILL TAKE TO BUILD...

If adapting for those small screens is a concern also, then much better would be to make Plasma Active around SVG... gazillion better for the looks of tiny things... and the windowing system is orthogonal in here...

Don't worry... HAVE VISION... don't know what some witchdoctors try to push, but the notion of ARM underpowered can only be laughable (BIG LOLzz)... YES MORE POWERFUL THAN SOME EVEN INTEL CURRENT i7 TOP DESKTOP VERSIONS... and next crop of GPGPUs in mobile could easily handle SVG allover, AND better(yes dare to say this) THAN the majority of current mainstream Desktops (remember AMD/ATI is on the boat now, nvidia tegra can't afford to make underpowered things anymore, PowerVR next GPUs are positioned to be 6x more powerful, ARM MALI T6## is about 5x more powerful)...

Why in the HELL anything that is in current Linux desktops, is in need to be adapted for Mobile (BIG ??)... if you are running absolutely *OBSOLETE* hardware then yes... but you can't expect to have success with those can you (example, the KDE tablets look very good, yet the hardware is already obsolete even before debut ARMv9!? -> crazy! ) ???

So to me what is missing is "COMPUTE POWER" and SVG (which could go hand in hand)... Wayland ?? ( doesn't solve anything just for starters...)

... step outside off the box they want to put you, smell the fresh air... you'll know its truth!!... how easy is to make things more powerful than with crap old x86, and how GPGPUs will be no more devices but POWERFUL PROCESSORS, MORE POWERFUL THAN CPUS (stay tunned for the next console games, their "seem" diminute specs, but how they can blow you away)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 3:39 UTC (Wed) by mmarq (guest, #2332) [Link] (1 responses)

Ooops exageration! (well in reference design is, others may excel)

ARM-A57 is not 5x more powerfull than the A15... the 5x applies only for the top GPGPUs T6## compared with T5##...

Concerning 32 bit (which the A57 also does, very identical to x86_64 in this)

http://translate.googleusercontent.com/translate_c?depth=...

The A-57 is around 2.25X more powerful than the A9...
The A15 which has a pipeline identical to A57 is around 2x more powerful than the A9 and 25% less powerful than the A57 -> BUT IS LESS POWER EFFICIENT...

Hey!... i'm sure 64 bit and it could be at least 50% difference in favor A57 compared with A15 for the same code(which is quite remarkable) ... also the ARMv8 is a "clustered" architecture, meaning it will be possible to have more than 4 cores in a single chip... so it may end up being 5x (add more cores) lol...

See for yourself (correct me please)
http://translate.googleusercontent.com/translate_c?depth=...

Family of GPU roadmap "Mali-T6xx" made by ARM to support Google Nexus 10
http://translate.googleusercontent.com/translate_c?depth=...

PowerVR "Rogue" 100GFLOPS stand next to raise the graphics performance of the smartphone and tablet
http://translate.googleusercontent.com/translate_c?depth=...

C'mon KDE!... tablets with obsolete A9, and ridiculous GPU... is that why wayland is needed ??

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 7:30 UTC (Wed) by mmarq (guest, #2332) [Link]

The writing is on the wall!

Tegra 4 will also be 6x more performance(like PowerVR) on the GPGPU side than Tegra 3.

http://translate.googleusercontent.com/translate_c?depth=...

From 12 ALUs to 72 ALUs... and the CPU side also will have more than double performance( it was if not wrong) 2 ARM A15 cores + 1 ARM A9 big.LITTLE. NOW its 4 A-57 cores + 1 A53 big.LITTLE... smaller process more efficient, and the A57 is more power efficient and the A15, and the A53 is *HALF* the power of A9, so it could be basically the same power envelope.

http://translate.googleusercontent.com/translate_c?depth=...

KDE... that is a chip to should be on those tablets...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 4:27 UTC (Wed) by mmarq (guest, #2332) [Link] (13 responses)

"" Eh!!!... and how do you change ~ 200.000 applications from any ties to X to be "tied" with Wayland ???...

As i said, X without network is not X ...

Wayland with X is just "another" X ...

Wayland is absolutely a gargantuan enterprise... to change an enormous app base to its side... ""

And i'm starting from the principle that indeed Wayland is better and faster than Xorg (and i say xorg so that you can think of "others" )... no problem at all with that...

The problem is that to support the *BIG LACKING* of every app being X, i'm afraid running an Xorg(or even other) server on top of a wayland server is going to make things slower than running that apps directly on even Xorg(nevermind others that most where already faster)...

This is almost a "blind" easy bet to do... that shadows all Wayland claims...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 5:16 UTC (Wed) by viro (subscriber, #7872) [Link] (12 responses)

You know, I'm not too fond of Wayland (and attitude re network transparency is a large part of it), but the only thing you are succeeding at is sounding like a kook. Talking to yourself, ALL CAPS all over the place, multiple exclamation signs, conspiracy theories... the only thing missing (AFAICS) is active sock-puppetry. Splashsnot is -> that way...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 8:57 UTC (Wed) by mmarq (guest, #2332) [Link] (8 responses)

Hey!.. at least i got your attention, which his not bad.. or is it ?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 9:35 UTC (Wed) by anselm (subscriber, #2796) [Link]

Hey!.. at least i got your attention

Yes, but not in a good way.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 12:52 UTC (Wed) by renox (guest, #23785) [Link] (6 responses)

For me, only this time: welcome to my filter list.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 22:37 UTC (Wed) by mmarq (guest, #2332) [Link] (2 responses)

Nice feature that filter button... but exactly as i expose just below, with proper training the scroll wheel can be as fast as the filter button...and you don't have to plug off the eyes the hears or the mouth...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 21, 2013 11:17 UTC (Thu) by anselm (subscriber, #2796) [Link]

The filter list is a good way of avoiding specific users who flood the comments to an article with humongous amounts of content-free stuff. I'd much rather not have to replace my mouse every two weeks because the scroll wheel is worn out.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 23, 2013 2:46 UTC (Sat) by viro (subscriber, #7872) [Link]

> Nice feature that filter button... but exactly as i expose just below

Gives a new meaning (hopefully) to the expression "indecent exposure"...

> ..and you don't have to plug off the eyes the hears or the mouth...

No comments.

Seriously, though, skipping occasional garbage by scrolling past it makes no sense when the source of that garbage is a twit of your caliber.

*plonk*

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 22, 2013 8:57 UTC (Fri) by loonyphoenix (guest, #89514) [Link] (2 responses)

How do I do that? I registered right now with the sole purpose of blocking that user, and I still can't see the button.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 22, 2013 9:00 UTC (Fri) by anselm (subscriber, #2796) [Link]

See »Comment Filtering« under »My Account«.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 22, 2013 9:15 UTC (Fri) by paulj (subscriber, #341) [Link]

You have to be subscribed at at least the "professional hackers" level for filtering to be enabled on your account though, as I understand it. If you are, but don't see the option, check that filtering is enabled in your preferences:

https://lwn.net/MyAccount/YOURUSERNAMEHERE/preferences/

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 9:17 UTC (Wed) by mmarq (guest, #2332) [Link] (1 responses)

And you know why... usually i read 100 posts in 2 or 3 minutes or less, its called "diagonal reading" meaning i don't read not even half of all written words, only read full posts if that are any "word or phrase that catches my attention"...

Sorry, i shouldn't be selfish thinking everybody does the same...

but there is also a reason for it, some most popular forums can have topics with thousands of posts trough hundreds of pages... and immense whole of garbage in the mix... so its not kook, its a style... but if you want call it kook i'm pretty fine with that... gazillion better a "kook" than a false person or a hired shill... what should matter is the truth...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 25, 2013 16:30 UTC (Mon) by nix (subscriber, #2304) [Link]

If 'what should matter is the truth', it's amazing you don't bother to do any research before posting your conspiracy-laden blitherings. Virtually everything you have posted on this subject is either irrelevant (e.g. the arrival of new smartphones, wtf does that have to do with X vs Wayland?) or provably false because self-contradictory (the ridiculous conspiracy theories which would require several core X and Wayland developers to be conspiring against themselves).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 25, 2013 16:27 UTC (Mon) by nix (subscriber, #2304) [Link]

I didn't know you could *do* green ink on LWN, but mmarq has found a way. I don't really understand how the heck the Wayland developers could be conspiring against the original X developers when many of them are the same people... split personalities, perhaps?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 20, 2013 22:51 UTC (Wed) by Serge (guest, #84957) [Link] (1 responses)

> If wayland will be such upper duper it will win without much effort... if not, just history repeating itself...

If only things were that simple...

The problem is that Wayland can win even being much worse than X.Org. Remember OS/2, "a better DOS than DOS and a better Windows than Windows"? Windows was no better than OS/2, still it won. :)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 21, 2013 8:49 UTC (Thu) by khim (subscriber, #9252) [Link]

Actually Windows OS/2 popularity grew significantly when OS/2 3.0 Warp was released (especially in some countries like Germany). But then contract expired and IBM had no right to create OS/2 for Windows 95 which sealed it's fate.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 16, 2013 0:08 UTC (Sat) by dlang (guest, #313) [Link] (6 responses)

> the code is PRE-ALPHA STATE.

This is the core of the problem.

It makes news when a distro decides NOT to make Wayland the default in their next (6 month out) release.

You have many advocates for Wayland effectivly claiming that anyone who doesn't agree that Wayland isn't AS OF RIGHT NOW the best thing ever are either stick-in-the-muds or a fringe element that should be ignored and not supported in the future.

These don't jive with "PRE-ALPHA state"

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 16, 2013 0:20 UTC (Sat) by raven667 (subscriber, #5198) [Link] (2 responses)

> You have many advocates for Wayland effectivly claiming that anyone who doesn't agree that Wayland isn't AS OF RIGHT NOW the best thing ever are either stick-in-the-muds or a fringe element that should be ignored and not supported in the future.

I don't think that's true at all, I haven't seen anything like that here, and I'm obsessive about reading lwn comments 8-). There are plenty of people who think wayland is the right direction for X to go in but that it is a work in progress. It'll probably be years before the major distros switch, but it will happen, there are too many benefits to not switch.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 18, 2013 23:15 UTC (Mon) by mmarq (guest, #2332) [Link] (1 responses)

That if it is NOT Open Source...

X without network is not X
Wayland with the X network protocol, is just X

So Wayland so far IS NOT an evolution of X, its a replacement, more centered around the "mobile" side(superphones/tablets and such)... and some how the agenda is against the possibility of OpenVPN home, and go by there virtue of X(something similar)... this interests going behind wayland want all going true the web(no private network for the desktop, only the public be available).

If Wayland is F/OSS, it might be that a couple of crazy gits decide to include the pertinent features in X itself... and you know what!?... this X system is faster and better in all aspects than Wayland...

Nothing particularly against wayland... go ahead please... but don't try to sabotage anyone else...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 25, 2013 16:35 UTC (Mon) by nix (subscriber, #2304) [Link]

more centered around the "mobile" side(superphones/tablets and such)

Absolute rubbish. Wayland is designed around the requirements of modern graphics hardware, and around the ways in which usage of X has changed over the years -- unless you think that all usage of Gtk and Qt is on smartphones and tablets, that is (which would, of course, be wrong).

If Wayland is F/OSS

If? If? You haven't even done that much research? Googling for 'Wayland', clicking on the first link, and clicking on 'Wayland FAQ' tells you as much in the second FAQ item.

Sheesh.

a couple of crazy gits decide to include the pertinent features in X itself

Given that most of the 'pertinent features' consist of getting X out of the way, something you can't do to an X server and still call it an X server, I'd say that this feature is self-contradictory (or, alternatively, has already been done: we call it DRI.)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 17, 2013 15:21 UTC (Sun) by renox (guest, #23785) [Link] (2 responses)

Bah that's why vocal "fans" must be ignored be they Wayland-fan or X-fan or whatever-fan.
This doesn't make Wayland less interesting..

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 17, 2013 19:30 UTC (Sun) by dlang (guest, #313) [Link]

the problem is that it's hard or impossible to tell the difference between 'fans who should be ignored' and 'developers who you should try and convince or you will not have the equivalent to a feature you use now'

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 18, 2013 23:30 UTC (Mon) by mmarq (guest, #2332) [Link]

who in their right mind would care ????...

Don't ignore wayland... but just give the possibility of the user to change to the windowing system they like... no one minimally sane is going to argue about this or with anyone that chooses not to use wayland...

Them lets see who wins the preferences... or is Linux desktop to be transformed into windoze ?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 8, 2013 16:33 UTC (Fri) by theophrastus (guest, #80847) [Link]

Are "decorations" traditionally, or required to be, limited to unitask frame buttons? (resize, move, minimize, close...)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 0:27 UTC (Sat) by mordae (guest, #54701) [Link] (3 responses)

Since N900 and GNOME 3 both came up with expose function where windows get little Xes for closing, I am looking forward to removing the top portion of decorations completely and getting more space. Is that sane?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 3:56 UTC (Sat) by Sho (guest, #8956) [Link] (2 responses)

You can configure that in KDE right now (the standard window decoration allows you to reduce the title bar to a pure border, and the Expose-like Present Windows effect has window close button overlays on the scaled windows). Example: http://wstaw.org/m/2013/02/09/plasma-desktopmh1137.png

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 3:57 UTC (Sat) by Sho (guest, #8956) [Link] (1 responses)

I should add that the Oxygen style engine also enables a "drag windows from empty areas" behavior by default, which comes in handy in this context (e.g. you can drag that window from the empty space to the left of the menu bar items or the status bar and similar, although Alt+Drag works anywhere in the window anyway of course).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 9, 2013 10:41 UTC (Sat) by mgraesslin (guest, #78959) [Link]

and if you use the option "No Side Border" or "No Border" a resize overlay is added (since 4.10), so that you can still resize the window without needing a resize button. Of course Alt+RMB also works.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 10, 2013 15:39 UTC (Sun) by Jandar (subscriber, #85683) [Link] (9 responses)

Now I look forward to a distribution having KDE on Wayland. CSD would be a sure deterrent but with SSD I'm eager to give Wayland a try.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 4:00 UTC (Thu) by Serge (guest, #84957) [Link] (8 responses)

> Now I look forward to a distribution having KDE on Wayland.

I guess it's not possible. You may get Wayland in KDE, i.e. KWin from KDE running in X.Org and supporting native Wayland applications. But it's really hard to run KDE on Wayland without X.Org.

Wayland protocol requires compositor to be a bloat monster. To have Wayland-KDE without X.Org you need KWin + Plasma + KScreensaver integrated into a single compositing application.

Are you sure you want that? ;)

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 9:34 UTC (Thu) by mgraesslin (guest, #78959) [Link] (7 responses)

> Are you sure you want that? ;)
No and that's why we won't do it. KScreenSaver is already dead in 4.10 (kept alive as a zombie for those who really need fancy animations from the 90). Plasma and KWin won't be merged and even in a Wayland world they don't need to be merged.

Sure more things will go into KWin, but that's not a particular new thing. It happened over the last few years more and more that we moved things into the compositor.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 10:56 UTC (Thu) by Serge (guest, #84957) [Link] (6 responses)

> KScreenSaver is already dead in 4.10 (kept alive as a zombie for those who really need fancy animations from the 90).

Different people, different tastes. On a private notebook screensavers may be useless. But I've often seen password-protected screensavers in offices.

BTW, have you seen rss-glx or Electric Sheep?

> Plasma and KWin won't be merged and even in a Wayland world they don't need to be merged.

In a Wayland world how can you implement taskbar, systray etc. without either merging plasma into compositor or extending Wayland protocol with the missing parts of X11?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 12:06 UTC (Thu) by hummassa (guest, #307) [Link]

IIRC taskbar, systray &c are NOT part of the X11 protocol, but extensions to it...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 13:07 UTC (Thu) by mgraesslin (guest, #78959) [Link]

>> KScreenSaver is already dead in 4.10 (kept alive as a zombie for those who really need fancy animations from the 90).

>Different people, different tastes. On a private notebook screensavers may be useless. But I've often seen password-protected screensavers in offices.
It's still possible to lock the screen, just KScreenSaver is not used for that any more.

>> Plasma and KWin won't be merged and even in a Wayland world they don't need to be merged.

>In a Wayland world how can you implement taskbar, systray etc. without either merging plasma into compositor or extending Wayland protocol with the missing parts of X11?
Systray is already D-Bus only (modulo legacy X11embedd and we won't have a wembedd) and for tasks I don't see a problem with having a private protocol between plasma and KWin if that doesn't get standardized. It's not like KWin and Plasma talk quite a lot already with custom protocols in the X11 world.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 15:51 UTC (Thu) by Kit (guest, #55925) [Link] (3 responses)

>Different people, different tastes. On a private notebook screensavers may be useless.
> But I've often seen password-protected screensavers in offices.

Personally, I have very little trust in the ability of X11 environments to securely lock a session. I remember for many years how windows would pop /over/ the screen saver. Or the screen saver crashing, defeating the security it "provided". Or the screen saver taking 30+ seconds to pop up after resume (why is it so slow to start???). Not to mention, under many circumstances the screen saver will never start (one of the Wayland videos from Linux.conf.au actually mentioned this).

Work has been done to improve the situation (I believe an extension was developed to try to solve the first one), but I still find the whole screensaver-as-locking-mechanism to be a poor fit. I'd rather see Linux adopt something akin to what OSX and Windows have been doing for somewhere around a decade, where the lock screen is its own secure context, instead of just a window painted above all the others. I believe the Wayland developers have even spoken about doing something along these lines in the past.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 14, 2013 16:55 UTC (Thu) by Cyberax (✭ supporter ✭, #52523) [Link]

Heck, I remember a bug in Compiz when you could just drag screensaver away using the alt-drag feature.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 10:26 UTC (Fri) by paulj (subscriber, #341) [Link] (1 responses)

A bug I've noticed with GNOME3 (GNOME-shell) is that the screensaver locking often hasn't taken effect when the machine suspends. So when the machine resumes, you are shown the desktop that was there before, and *then* (sometimes after seconds) it switches to the lock screen. This could be a problem if there was sensitive content on the desktop prior to suspending.

I guess the GNOME3 screen saver isn't synchronised with the suspend process, such that it ensures the screen locking code has run before suspend occurs.

I don't remember either way whether or not this occurred before GNOME3. Maybe it's a long-standing bug, maybe not. If a long-standing Linux suspend / screensaver bug, then I suspect it's more noticeable with GNOME3 because of some slowness somewhere or races that become more apparent with the heavier-weight nature of GNOME shell (RAM wise particularly) - resume takes a lot longer with GNOME shell.

On my multi-monitor, work desktop, I've noticed with GNOME3/Cinnamon the screensaver can sometimes kick in but still leave 1 desktop monitor setup visible - though you can't interact with it. You have to type the password blind to get in. Fun...

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 10:28 UTC (Fri) by khim (subscriber, #9252) [Link]

It happened before GNOME3, too - it just was less annoying back then.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 15, 2013 21:45 UTC (Fri) by Serge (guest, #84957) [Link] (3 responses)

As for why wayland was initially CSD-oriented, I guess, it's because this is easier for compositor. Server side decorations would require a lot of additional work: make a way to change title, configure supported buttons, borderless popups, window resizability... oh, let's just push those things to toolkit and let them worry about that. More work for toolkit, less work for compositor. Compositor authors should obviously want to make their work easier. :)

I wonder how wayland is going to allow sane tiling with client side decoration. IMHO, it's either good tiling support or CSD.

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 17, 2013 15:17 UTC (Sun) by renox (guest, #23785) [Link] (2 responses)

> As for why wayland was initially CSD-oriented

Initially? The Wayland dev are still for CSD.

> I wonder how wayland is going to allow sane tiling with client side decoration. IMHO, it's either good tiling support or CSD.

I'm not sure that they care much about tiling, but what is exactly the issue with CSD and "good tiling"?
If the compositor can tell the clients "don't draw your decorations" or "draw your decorations", wouldn't this be enough for "good tiling" and CSD?

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 17, 2013 23:18 UTC (Sun) by Serge (guest, #84957) [Link] (1 responses)

> Initially? The Wayland dev are still for CSD.

IIRC, implementation of wayland in kwin used server-side decorations from the very beginning.

> what is exactly the issue with CSD and "good tiling"? If the compositor can tell the clients "don't draw your decorations" or "draw your decorations", wouldn't this be enough for "good tiling" and CSD?

Yes, that would be the beginning. But wayland compositor can't do even that. It also does not know about window titles. For the same reason it's kind of impossible to implement a taskbar/dockbar in wayland (no information about window title, icon, etc).

Gräßlin: Client Side Window Decorations and Wayland

Posted Feb 18, 2013 10:01 UTC (Mon) by renox (guest, #23785) [Link]

>> Initially? The Wayland dev are still for CSD.
> IIRC, implementation of wayland in kwin used server-side decorations from the very beginning.

Sure but KWin isn't made by Wayland developers, it use Wayland which is different, Weston made in the Wayland project is still CSD only: the Wayland protocol supports both type of design.

> But wayland compositor can't do even that.

Weston cannot do it currently which isn't a big issue: the main issue is that if this 'decoration negotiation' part isn't part of the basic protocol (I don't know if it is or not: I haven't checked) then this will make interoperability complicated.