mozilla :: #firefox

11 Sep 2017
00:32Zinkerhi
00:33Zinkeris there an alternative h264 codec for html5
00:35Caspy7Zinker: Firefox relies on the platform for the codec. On Windows it can only use the built-in codec. On linux it uses ffmpeg
00:37Zinkeric , i have codec version 1.6 and read theres 1.7 ? update button no effect
00:40Caspy7Zinker: you have not stated what your issue is. Is it not working?
00:40Caspy7and I just assume we're talking about ffmpeg?
00:41Zinkerhmmm
00:42Zinkerwho can i read the coed is used by a html5 player ?
00:42Caspy7I do not understand the question
00:42Zinkeron a specific site
00:43azakiffmpeg is at 3.3.3 right now.
00:43Zinkeri have the idea different sites use different player and codecs , is there a way to xypher it ?
00:44azakimost websites use h264
00:44azakigoogle also offers vp9
00:45Zinkerlike this , yes
00:45azakias for different players.. eh.. i guess.. the player widgets is made in a mix of web technologies..
00:45Zinkerit says only htl5 player stabile 3.12.x
00:45azakiuh =o
00:46Zinkeric hmm
00:46Zinkerthis makes it more complicated , widgets
00:47Zinkeri must think over that a few days
00:47azakiZinker: i don't even know what you're saying or asking...
00:47azakior what your problem is.
00:49Zinkeravc1.42E91E what codec is that
00:49Caspy7azaki: oh, I see you've arrived to where I am :)
00:49Zinkeri cant copy and paste in this window ..
00:49azakiCaspy7: I was trying my advanced decryption ciphers on their statements.
00:49azaki=p
00:50Caspy7:)
00:50Zinkeravc1.42E01E what codec is that
00:50Zinkeryaya funny yaya
00:51Zinkerwhat codec is that :)
00:51azakiaccording to google "AVC1" is how some players refer to H264. =o
00:51azaki'AVC' is another well known name for H264.
00:52Zinkerwindows codec , cisco codec , own written codec , widget codec ?
00:52azaki...
00:53Zinkeraha , now all understand me
00:53Caspy7meh
00:53azakiH264/AVC is a codec standardized by the MPEG-LA group (and patent pool)
00:54Zinkeroffers who?
00:54Zinkercome ya did understand well , ya dont know also
00:55Zinkeri wantto update that codec , how
00:55Zinkerim on windows
00:57azakiyou can't. it's bundled with windows.
00:57azakiif you think you need to 'update it', something else is wrong.
00:57Zinkerwell its with media player ...
00:58azakiif video isn't playing on a web page, you might try disabling your adblocker or some other addons that could block stuff on the page.
00:58Zinkerits palying , but stutter
00:58azakiget a faster computer? =o
00:58azakior maybe install linux.
00:58azakiwhich uses less system resources than windows
00:59Zinkerccou usage 1 %
00:59Zinkernot sure its the computer
00:59azakithen maybe your browser has too many tabs or addons installed. firefox can often get sluggish in those circumstances
00:59Caspy7Zinker: Try testing in safe mode (Help > Restart with addons disabled) This will also disable hardware acceleration
01:02Zinkertrying safe mode
01:02Zinkerbut no way to trick firefox to use the coedec i want ?
01:03Zinkerin windows
01:03azakii don't think the problem is the codec... otherwise plenty of others would be having the same issue
01:03Zinkerx-mpegUrl or so , i cant paste ...
01:04Caspy7Zinker: most websites don't use alternative codecs except youtube
01:04Zinkeric
01:04Zinkeri must test safemode for a day to make statement
01:06Zinkeridea i google windows update h264 codec
01:07Caspy7Zinker: if it's a hardware acceleration issue, then you can test with hardware acceleration disabled (requires a browser restart) but the best resolution for that is to ensure your GPU drivers are completely up to date
01:07Caspy7What version of Windows is this?
01:07Zinkerwindows 7 with all updates
01:09Zinkermaybe my vld player is too old , this is odd
01:09Caspy7Zinker: if it has all updates then your H264 codec is completely up to date
01:10Zinkeri thought also so ...
01:10Zinkeri google what version i have
01:11Zinkerwindows 7 h264 codec version find out anyone a better stiring ?
01:13solenodicI'm curious
01:13solenodicwhat's going to happen to userchrome.css as Quantum progresses
01:14solenodicwill the next firefox UI be written using UI abstractions that are as stylable as XUL firefox is?
01:16solenodicmaybe some abstraction over webrender that provides finer grain control but can be styled using CSS and the DOM inspector
01:17Caspy7solenodic: nothing is going to happen to it
01:17solenodicsurely some XUL cruft being removed will change the way it works, though
01:17solenodicor will it try to be relatively backwards compat?
01:18Caspy7solenodic: XUL is a combo of HTML + JS + CSS and custom stuff. The custom stuff will eventually get removed
01:18Caspy7the the CSS won't
01:18Caspy7*but the
01:19solenodicso my understanding
01:19Zinkerok i try a day safe mode ,back tomorrow
01:19azakimy understanding was that long-term the goal is to eventually remove XUL entirely
01:19solenodicthere'll be a rust representation of the UI that exposes its inner parts to CSS code
01:20solenodicrather than an XML structure to represent the UI
01:20azakiand that the UI will be pure HTML5 or something
01:20Caspy7azaki: yes, then it will be HTML, CSS & JS
01:20Caspy7solenodic: ^
01:20solenodicsounds like the best way to go about it
01:21azakii'm curious if they will start rendering page elements like scrollbars and buttons and stuff themselves instead of using the system toolkit.
01:21solenodicwill the javascript in the UI access significantly different APIs from the standard HTML5 apis
01:21azakii think blink has already gone in that direction
01:22azakithis would minimize the dependency on system toolkits to just being used for print dialogs and so on, and would increase the amount of code shared between platforms.
01:23azakiand thus remove a lot of the complexity in maintaining an abstraction for different platform toolkits
01:25solenodiconce quantum is complete it seems like a good opportunity to bring back support for some "privileged" extensions
01:25solenodicthat has DOM level control of firefox's chrome
01:25azakii don't think they will bring back things that can't be easily made backwards-compatible.
01:26azakione of the goals in the new extension system is to keep things compatible.
01:26solenodicwebextensions are great for 99% of things you need extensions for, but you can't really get a more capable API than hacking right into the inner representation of the UI
01:26solenodicI would probably agree that support for these "privileged" extensions should be a compile time flag though
01:27solenodicwe're just now starting to prevent average joes from installing malware in their browsers
01:27azakiat that point why not just use a webext experiment in an unbranded build? =p
01:28solenodichow living is the webextension standard anyways?
01:28Caspy7solenodic: part of the point of the new system is to have proper APIs for the UI because the old system allowed addons to break the browser easily. Having DOM access sounds like the same thing as the old.
01:28Daggeryou should be able to sign the extension, and the system adminstrator should be able to add accepted signing roots for loading extensions
01:29Dagger(and really it should be possible to remove Mozilla's root too, in case you don't trust them and want to exert your own control over what runs. similarly you should be able to add "app stores" other than addons.mozilla.org)
01:30solenodicdoes firefox support sticking in your own signing keys without compiling and making an unbranded version?
01:30Daggerwe're Mozilla. we're supposed to be more like Linux, not Apple
01:30azakithat might be reasonable for 'developer edition', but i would say it's unreasonable for the release version.
01:30Daggersolenodic: no, it doesn't. which is something we really should've fixed ages ago
01:31Daggerazaki: release version is what we expect people to run, so obviously it needs to work there too
01:31Daggerwhat's the point of an admin control that doesn't work?
01:31solenodicalthough, how would you securely do that so that malware can't sign its own malware extensions?
01:32Daggeryou put the config in /etc, where only root can change it
01:33solenodicthat's great for us unix users
01:33solenodicbut windows users are often way more lax about securing administrator access
01:33solenodicsome users still use the admin account all the time...
01:34Daggerif that's how they want to run their system, they're free to do so
01:34solenodicpeople're still using windows XP :(
01:35Dagger<--
01:35Daggernot running as Administrator though
01:35solenodicI just think that the &quot;average people version&quot; should probably defend itself against all the malware the average person can install on their computer
01:36solenodicI don&#39;t think just a &quot;developer edition&quot; is necessary though
01:36solenodicdeveloper editions are for developers
01:36MardegDagger: are you forgetting Slacko Puppy runs as root? :P
01:36solenodicmaybe an &quot;OEM&quot; edition
01:36azakiDagger: the point is the browser has to work to protect most users. if most users are dumb in regards to security, the the browser has to assume lowest common denominator.
01:36solenodicbut not in the traditional sense of an OEM
01:36solenodicI&#39;m thinking more of distributions of free software
01:36azakihence why it&#39;s more reasonable to have such advanced functionality in a dev edition
01:37solenodicDagger: you are using winxp?!?
01:37solenodicdaily driver?
01:38Daggeryep. what else am I supposed to use?
01:38solenodicwindows vista obviously
01:38MardegLOL
01:38DaggerWin10 is Win10. you can&#39;t tell me &quot;Win7&quot; because that&#39;s out of date too...
01:38solenodicunfortunately windows vista died 2017 april 11
01:38Daggerobviously I should be on Linux, and I&#39;m working on that, but ugh, Linux GUI
01:39solenodicbut windows 7 isn&#39;t &quot;out of date&quot; in the same sense as winXP is
01:39azakiyou&#39;re shitting me, you&#39;re really on xp? lol
01:39solenodicwindows 7 will receive at least some security updates
01:39Daggerthere are still Linux toolkits that can&#39;t handle multiple monitors with different y heights properly
01:39azakii suddenly feel very different about our various debates... =p
01:39solenodicwindows XP will not receive any security updates
01:39Daggerspecifically you try to open a menu, and they render it a few inches away from the menu bar
01:39solenodicI&#39;m hoping they&#39;re not gonna make any more exceptions to no updates for XP
01:40Caspy7azaki: I was thinking the same thing
01:41solenodicthe risk of really bad ransomware attacks should be an indicator to anyone running large institutions on windows XP that they should transition away from it ASAP
01:41Daggerazaki: yeah, I don&#39;t exactly like the situation, but it works for me and everything else doesn&#39;t
01:42Daggersolenodic: I mean, I do have backups and automatic snapshots, I should be able to roll back something like that
01:42azakiDagger: yeah, does everything else &quot;break your workflow&quot;? =p
01:42solenodicwhen you keep giving them updates they have no reason to worry
01:42Daggerat least... for some of my storage. some of my storage drives are still NTFS :/
01:43azakiI&#39;m not aware of this toolkit menu issue you&#39;re referring to though.. I&#39;d need a screenshot. Unless you&#39;re talking about some obscure toolkit that isn&#39;t Qt or GTK ...
01:43azakilike fltk or fox or something..
01:43Daggerazaki: looks ugly, treats me like an idiot, or just plain doesn&#39;t work right
01:43solenodicDagger: roll back after your privatest data is stolen, including your hard drive encryption key?
01:44solenodicor &quot;nothing to hide&quot;?
01:44DaggerI had to spend some time getting pcmanfm to handle the mouse back/forward buttons, for instance. it just interprets them as left-clicks otherwise. Linux is full of stuff like that
01:44azakia lot of gui issues can be traced back to issues with X11 though in some cases.
01:45azakiwell... if you use an obscure file manager like that... >_>
01:45Daggerazaki: I had the menu problem with VirtualBox at one point, so not exactly obscure. I can&#39;t remember what programs have the problem on my current Linux machine though
01:46azakione of the reasons i use gnome is precisely to avoid all the annoying little quirks of smaller DEs. the issue is that linux itself has little desktop marketshare, but within that, the small DEs have even *less* .. which means they may not get bug reports for really common things because the devs are all like &quot;works for me&quot;
01:46Daggersolenodic: well that wouldn&#39;t be ransomware
01:47azakione of the reasons i&#39;m glad that ubuntu is switching back to gnome is precisely that it will lead to more testing and bug reports.
01:47solenodicDagger: right, I&#39;m just saying. all kindsa malware can fsck your day up
01:47Daggerazaki: yeah, I&#39;m on XFCE. not GNOME, for obvious reasons
01:47azakii used to use xfce. that&#39;s what i switched from.
01:48azakixfce&#39;s compositor used to be completely broken, for like 3 years or something.. because they don&#39;t do major releases often.
01:48solenodicI half wish that wayland had taken over a few more roles
01:49solenodicthe compositor being able to be separate from the window manager is nice
01:49azakiif you enabled it, you eliminated application tearing when resizing and stuff, but you get video tearing when playing movies and shows..
01:49azakibut then if you disable compositing, you dont get video tearing..
01:49azakibut then the desktop tears. lol
01:49solenodicmaybe what I wish for
01:49azakionly solution was to use a standalone compositor..
01:50solenodicis a wayland-like thing that can run multiple compositors at the same time
01:50Daggerazaki: blargh, yes. I had so much trouble with tearing. I&#39;ve got something which mostly works, but now I&#39;m afraid to touch any of the settings
01:50azakisolenodic: in different TTYs? you can.
01:50solenodicI mean multiple things able to influence how windows are displayed to the screen
01:50azakiDagger: that was one reason i switched to gnome. it uses GL compositing which mostly works.
01:51azakithese tearing issues are mainly X problems.
01:51solenodiclike if you&#39;re using sway and you wanted to code in an animation that makes switching workspaces have a slide effect, then you could make that code separate
01:51azakisolenodic: why... you understand that compositors can still choose to split the responsibilities into separate apps right?
01:51solenodicthat code would be aware of i3&#39;s internal state just from its IPC
01:51solenodicsway* not i3
01:51azakithey could use dbus, or whatever
01:52azakithere is a kernel ipc being worked on called bus1
01:52solenodicazaki: then give me a scriptable compositor
01:52azakii dunno if that&#39;s tentative name or whatever, but yeah. it&#39;s designed to be super fast and have minimal copies. eventually they hope to get it merged into the kernel.
01:52Daggersolenodic: re:malware, you&#39;re right, of course. it&#39;s one of the reasons that I&#39;d like to finally get off XP. maybe one of these days...
01:52solenodicwell not a scriptable compositor
01:53azakisolenodic: well, they just decided not to take on those roles. someone else could design a specification to do more things, and split up the roles into different programs. and it would still be a wayland compositor.
01:53Daggeralthough I seem to be doing okay on the malware front -- but by saying that I&#39;ve probably just condemned myself to not doing okay
01:53azakii think it&#39;s a safe bet because if they&#39;d bitten off even more work, then it would&#39;ve taken even more time to ship something.
01:54azakiwayland is shipping in default now in fedora.
01:54solenodica compositor that can be controlled with a decently fast IPC interface
01:54azakiand will ship by default in ubuntu 17.10
01:54azakiwhich is the first version of ubuntu to go back to gnome
01:54solenodicthat handles basic animations
01:55azakithere are security concerns with making it scriptable, would have to be a privileged interface.
01:55azakisome people have been working out how to do privileged wayland APIs
01:55azakisince they need it for things like screen recording.
01:56azakiX has zero security
01:57solenodicso like
01:57solenodicyou could have APIs that the compositor tells wayland about
01:57solenodicand then those APIs can be exposed to certain programs you allow it to?
01:58azakitells wayland about ? =o
01:58azakithe compositor is wayland.
01:58solenodicright, it tells the wayland library
01:58azakior to be clear, wayland is two libraries that implement the IPC that compositors use.
01:59azakiyou mean a protocol extension?
01:59solenodicthat&#39;s probably a better term for it
01:59azakiyeah, that&#39;s one way to do it. but the compositor could expose such protocols using anything it wants to.
01:59azakiit could invent it&#39;s own protocol using unix domain sockets
01:59azakior use dbus
02:00solenodichowever if it used wayland&#39;s IPC
02:00azakior the new bus1 kernel ipc, which isn&#39;t merged yet.
02:00solenodicthen it would inherit wayland&#39;s security model
02:00azakior yes, it could use wayland protocol extension and go through wayland ipc.
02:00solenodicso that it&#39;s all standardized and neat and clean
02:00azakiwell.. wayland&#39;s security model depends on not exposing such functionality. the security is basically &quot;not providing functionality that is dangerous&quot; =o
02:01azakiin other words it&#39;s restricted by default
02:01azakiand then you need policy systems to enable privileged functionality
02:01azakilike for instance, you can already record the screen or take screenshots in gnome wayland. you just can&#39;t use third-party apps to do it yet.
02:01solenodicso wayland doesn&#39;t really have a capable policy system for doing that?
02:02solenodicit&#39;s just &quot;do you have permission to display a window or not&quot;
02:02solenodic?
02:02azakithis works securely because the compositor knows that the only way to emulate a mouse is to go through the kernel (which requires root access), or with a compositor plugin, which the compositor would obviously know about.
02:03azakiso no app can automatically click &#39;record&#39; or &#39;take screenshot&#39; for you.
02:03azakilike malware.
02:03azakiit knows that the user clicked the record button
02:03azakibecause the compositor has full knowledge of the GUI session and what is going on in there
02:04azakiso by building the screen record function into the compositor, that makes it secure
02:04azakibut if instead you&#39;re trying to trust a third-party screen recorder, like OBS
02:04azakiit&#39;s trickier because how does the compositor know that the user &quot;trusts&quot; that program?
02:05azakiif this was exposed as just an API that any program could use, then malware could use it too
02:05azakito record private info on your screen.
02:05azakiso some secure mechanism is needed to restrict apps that the user doesnt trust, from doing things like that. =o
02:06azakii&#39;ve been told it&#39;s possible with sandboxed apps like flatpaks.
02:06azakiso an OBS flatpak could be &quot;trusted&quot; for instance by wayland to do recording. =o
02:07solenodicI kind of feel like flatpaks and similar technology are ignoring what linux&#39;s users/groups are designed for
02:07azakisolenodic: the policies are done by you, the compositor writer, basically. the protocol is strict enough to prevent apps from knowing where other apps are on the screen, and what they are doing, and keeps them isolated and sandboxed from eachother (UI sandboxing, obviously at lower levels you need other ways to sandbox them, wayland just does the UI part for you)
02:08solenodicbut then again we&#39;ve already ignored that years ago
02:08azakialso, copy and paste protocols are designed so that explicit user action is needed, whereas in X any program can read the clipboard.
02:09solenodicalthough, was it nixOS that gives you the option to install the distribution&#39;s software on specific user accounts
02:09azakiso this is where the security comes from, in the sense that the protocol is designed with isolation in mind. but for the most part, compositor devs have to come up with the policy.
02:09solenodicrather than onto the whole system
02:09azakilinux&#39;s users/groups aren&#39;t really powerful enough for this. sandboxing is meant to solve a different problem.
02:10azakiif you run every program as it&#39;s own user or group, you&#39;ll just flood the system with users and groups =\
02:10* solenodic wants to see how the redox OS project tackles these problems
02:10azakithe issue is that you may not trust that program to have access to writing in your home directory for instance.
02:11Daggery&#39;know, I approve of UI separation et all, but if you want to run a program that monitors the clipboard for changes then you should be able to do that too
02:11solenodicDagger: yeah that&#39;s kind of what I say about firefox addons
02:11azakiwhen you think of it, the root/user separation on modern systems is fairly useless for many home users, given that most people care more about the content in their home directory, rather than the stuff in the system.
02:11azakithere was a steam bug years ago that deleted all your home directory =\
02:11solenodicprivileged addons would give choice to those downloading a customizable build
02:11azakithat&#39;s the kind of stuff that sandboxing and isolation prevents.
02:12Daggerazaki: what I really want is network isolation. I&#39;d like to be able to prevent a program from making any network connections, or control where it connects to
02:13azakiDagger: that is privileged behavior. you could write a compositor that exposes such an api, nothing in the wayland protocol prevents this, this is just how they&#39;ve designed the core protocol with isolation by default.
02:13azakibut you can always expose additional functionality if you write your own compositor, or modify existing ones, etc.
02:13Daggerif I don&#39;t want to give it access to /home then I can just run it as a separate user
02:13azakisounds like SELinux stuff.
02:13azakiSELinux has capabilities to restrict pretty much anything... including socket use by programs
02:13azakietc
02:14solenodicunfortunately SELinux isn&#39;t well supported in a lot of distros
02:14ali1234the problem is that OBS is not a screen recorder
02:14azakiit&#39;s way more advanced than say, apparmor.
02:14ali1234and duplicating all of its functions into every compositor is not an efficient use of developer time
02:14solenodicfor arch you really have to fuck with everything to get some of the linux security features to work well
02:14ali1234as a result, compositors will end up with a recording API
02:14ali1234and all the arguments about how wayland is trying to be secure will be forgotten
02:15ali1234it will be just as insecure as before
02:15azakiali1234: i think you missed the crux of what i was saying. i was using that example to illustrate how the isolation model works in wayland, if you scroll up i already mention that there is a *need* for a privileged API to support apps like OBS.
02:15ali1234but now we&#39;ll have 10 slightly different implementations of the same API that don&#39;t quite compatible with each other
02:15azakiyou are misunderstanding many things...
02:15solenodicali1234: why can&#39;t that compositor selectively allow certain software to access certain APIs
02:15azakisolenodic: it can, he/she is just confused
02:15ali1234i&#39;m not misunderstanding things, i&#39;ve argued for such APIs for years now
02:15solenodicmaybe even an android/webextension like permission system
02:15azakiand misunderstanding the issue/argument.
02:16ali1234wayland will never provide them, gnome will never provide them either
02:16azakiali1234: then stop spreading FUD and saying it will be &quot;just as insecure as before&quot;
02:16solenodicwhere the software could prompt you whenever another another program requests to record the screen
02:16ali1234solenodic: someone proposed such an extensionand wayland devs rejected it
02:16solenodicand remember those policies
02:16azakiyou don&#39;t know that actually... in fact, they&#39;re already working on something called PipeWire that will allow this. (red hat is)
02:16Daggerazaki: maaaybe. I think there was some problem with SELinux
02:16azakialthough pipewire has a larger scope than just this.
02:16ali1234that was about four years ago now
02:17solenodicali1234: I feel like it&#39;s appropriate for that to be part of the compositor and not wayland though
02:17azakiDagger: the only problem is that it&#39;s hard to use. but no other MAC system is as powerful/capable. all of the complaints about SELinux is basically that it works &quot;too well&quot; i guess. =p
02:17azakias in, it blocks things.
02:17Daggerin fact I think there were a lot of problems with SELinux. though perhaps I should investigate more
02:17azakithat&#39;s usually just how it&#39;s configured though on distributions
02:17ali1234yeah, just like how html implementation details are up to the browser
02:18ali1234that is totally a great idea, doesn&#39;t cause problems at all
02:18azakiyou can make it more permissive though with configuration
02:18solenodicselinux works too well? it sounds like it&#39;s for me
02:19azakisolenodic: I don&#39;t know where ali1234 is getting their information, but wayland did not &quot;reject&quot; a system like android... in fact, it was already proposed by gnome devs years ago.
02:19azakithey call them &#39;portals&#39; in gnome
02:19ali1234azaki: https://www.x.org/wiki/Events/XDC2014/XDC2014DodierPeresSecurity/xorg-talk.pdf
02:20azakisolenodic: people find it annoying because they&#39;ll run an app and it wont be able to do things because of SELinux policy, and then they&#39;ll be angry.
02:20azakiali1234: i think i&#39;ve seen this idea before, i dont think this is like android.
02:20azakithis was a different proposal
02:20ali1234azaki: this proposal was outright rejected by wayland, specifically by those wayland developers who also work on gnome
02:21ali1234and even more specifically those developers who work on gnome/wayland integration for red hat
02:21azakithey haven&#39;t rejected having a screen recording api though... as i said, they&#39;re already working on something called pipewire that will allow this...
02:22ali1234great. that still leaves about 100 other things that are not covered
02:22azakithis proposal had issues as i recall. i think it identifies programs by paths or something
02:22azakirather than a more secure mechanism
02:23azakias i recall the consensus is that whitelisting flatpaks for privileged APIs would be easier
02:23ali1234btw, WSM is designed like the new android security model, where it asks efore using a feature for the first time, rather than asking for everything at install time
02:23azakior any sandboxed format.
02:23azaki&quot;how&quot; does it ask though...
02:23azakiif it&#39;s like UAC prompts in windows
02:23azakithen that is really stupid and unworkable.
02:23ali1234it checks the installed policies
02:23azakiit can&#39;t be a dialog that the user will just click &quot;ok&quot; on autopilot.
02:24ali1234if there&#39;s no policy found then it is up to the compositor to decide what to do
02:24ali1234this is so distributions can pre-approve software
02:24azakithe way portals were designed for instance, users will not have an &quot;ok&quot; button to click, instead there will be a &quot;cancel&quot; button, and a choice of some sort, like if it&#39;s a webcam api, &quot;which camera to use&quot;, with a list.
02:24azakiwhich means you can&#39;t autopilot the choice
02:24azakiand have to actually read the dialog
02:24ali1234and if you want a copositor with UAC style prompts then you are free to make that
02:25azakiwhich is an important consideration.
02:25ali1234if you think you know a better way you can do that too
02:25azakithe better way requires a special api though, as in the example i just gave for instance.
02:25azakia generic api wouldn&#39;t let you create a list of choices like that for a specific use case like webcams
02:26azakiyou&#39;re trying to make something that is very complicated seem like it&#39;s easy
02:26ali1234this is actually a reasonable objection
02:26ali1234i&#39;ll be sure to mention it to the authors next time i see them :)
02:26ali1234i&#39;ve never heard it before
02:26ali1234it&#39;s certainly something that could be baked into a generic API though
02:27azakii&#39;m not sure but i may know them, the pdf uses the real name(s), i&#39;m more familiar with the irc names. if these people hang out on freenode in the wayland channel i may have talked to them already.
02:28ali1234sidi and mperes i think
02:30azakihm, maybe not then, i dunno. i do remember being linked to the wsm proposal on github, in the wayland channel
02:30ali1234yeah they have a github
02:30ali1234believe it or not i dont hate wayland
02:31ali1234the security stuff annoys me. of course security is easy if you just declare it Somebody Else&#39;s Problem
02:32azakiwell, to be specific, they declared how to bypass it to be someone else&#39;s problem. but i think it makes sense given that one of the main problems of X was it bit off too much, and thus has become difficult to replace due to it&#39;s monolithic design.
02:32azakii think keeping the scope narrow was one of the better things they did
02:33ali1234the main pronlem with X is all the extensions that are poorly defined
02:33azakibecause otherwise you get DEs to argue about the little details
02:33azakilike kde and gnome can&#39;t even agree on client-side decorations vs server-side decorations
02:33azaki=\
02:33ali1234if there was one true impllementation of NETWM, things would be much much better
02:34ali1234i see wayland ending up the same way if it forces everything to be an external extension
02:34azakiand there&#39;s also an issue with the xdg screensaver inhibitor spec where kde has decided to do their own thing, and gnome is following the xdg spec... and kde wont fix it..
02:34azakiso as a result, mpv cannot inhibit the screensaver on kde
02:34azaki>_>
02:34ali1234yes, and this is exactly the same kind of problem you&#39;ll get with wayland
02:34azakithey still haven&#39;t even agreed on xdg_shell yet.
02:35azakino, you&#39;re missing the point. wayland makes it easier to ship something, because you dont have to wait for everyone to agree.
02:35azakithe fact that gnome is shipping already actually does put pressure on kde to get their act together.
02:35azakiif you increase the scope of wayland
02:35azakiyou run into more &quot;differences of opinion&quot; between the DEs
02:36azakiand more bikeshedding
02:36azakimaking it impossible to ever ship the damn thing
02:36azakithe more control you give DEs, at least you can ship it. and then hopefully over time we&#39;ll see people agree to some freedesktop standards together
02:36azakilike they did with X
02:37azakinot sure how long you been using X, but kde and gnome used to have their own clipboard standards back in the 90s and even up to like early 2000s...
02:37azakithey also had their own application menu standards
02:37ali1234at the current rate that will take another 15 years at which point wayland will be exactly as bad and as exactly as old as X11 is now
02:38ali1234yeah X11 in 1996 was about as usable as wayland is today
02:38azakinah. wayland is fairly future proof, i mean it doesn&#39;t even have any rendering primitives. it&#39;s just buffer sharing and management.. that&#39;s kind of low hanging fruit.
02:39azakialso, no global coordinates, so even a vr 3d compositor would be possible.
02:39azakiwhich is a bit sci-fi ish, but eh.
02:39azakiwhatever
02:39azakithis is another benefit of keeping the scope minimal
02:40azakithe less responsibility you take on, the less of a chance you run into obsolescence as things change.
02:41azakii&#39;d say wayland is more usable than X in 1996. copy/paste and drag n drop work well in a standard way. =p
02:42ali1234but there&#39;s no drivers for my video card
02:42azakiwayland works on nvidia. =o
02:43azakion gnome that is.
02:43azakijust doesn&#39;t do Xwayland yet. ;p
02:43azakiwhich i guess is bad for non-GTK3 or non-QT5 apps. =p
02:44azakior apps that hook into X directly, like firefox probably does in various areas.
02:44ali1234it was extremely amusing watching nvidia drop a weston backend and then saying &quot;you designed it this way, you implement it in gnome&quot;
02:44azakifirefox is being ported to wayland as we speak though.
02:44azakia redhat dev is working on it.
02:46azakii&#39;m not sure what was amusing about that, nvidia is lucky that gnome were willing to negotiate. at first compositor devs were pretty much in agreement to snub nvidia&#39;s eglstreams spec. (nvidia decided not to use the standard gbm allocator in linux for their wayland drivers, they use eglstreams instead)
02:46azakibut after a bunch of discussion, it seems now everyone is working on a new allocator that will be better than both gbm and eglstreams =o
02:46ali1234its amusing because i told them years ago that this would happen
02:46azakiso yeah.. one more thing to wait for.
02:47azakitold who?
02:47azakithat what would happen specifically?
02:47azakignome only implemented eglstreams support under condition that the new allocator gets worked on.
02:47ali1234who&#39;s the guy who worked at collabora
02:47azakiso eventually everyone will be switching to that.
02:48azakidaniel stone ?
02:48ali1234that&#39;s the one
02:48ali1234remember when he wrote a weston backend for raspberry pi videocore overlays?
02:48ali1234then tried to use it as an example of how great wayland is
02:49ali1234that&#39;s when i pointed out that he&#39;d only demonstrated that every compositor must implement a backend for every video card
02:49ali1234and challenged him to compare wayland performance vs x11 performance on nvidia&#39;s binary driver
02:49ali1234funnily enough that rpi backend was deleted a couple of weeks after the eglstream stuff landed
02:50azakiwell, eh... if you read the deletion post, there was a good reason for it.
02:50ali1234there was a good reason for it: it made wayland devs look bad
02:50azakiit has bitrotted. the point of it was to show how flexible wayland is. but as the open drivers for raspberry pi mature, there&#39;s no point in it anymore
02:51ali1234so write a new one for eglstreams then?
02:51azakii think this is very conspiratorial..
02:51ali1234its not conspiratorial, its just people being lazy and declaring everything SEP.
02:52azakithe r-pi backend was useful as a demo to show how even on an underpowered platform, wayland can smoke x11. now that r-pi has open source drivers as part of mesa, and the r-pi backend has bitrotted, they removed it. it didn&#39;t make them &quot;look bad&quot; or anything
02:52azakiit just outlived it&#39;s purpose..
02:52azakiand eglstreams will eventually also be removed. =o
02:52azakionce the new allocator ships
02:53azakialthough it&#39;s possible rhel 8 may support it for awhile longer since it&#39;s enterprise, i guess we&#39;ll have to see how the timing lines up. i imagine rhel 8 will ship wayland support.
02:54azakialso nvidia is the only holdout, everything else uses the same backend.
02:54azakiand eventually so will nvidia
02:54azakithat was a mandatory condition in the agreement between nvidia and the gnome devs in regards to eglstreams support.
02:54ali1234as i told daniel stone several years ago. if you want to demonstrate that wayland is better than x11, do it using nvidia&#39;s driver, not a driver that isn&#39;t designed to run X11 at all
02:55azakinvidia didn&#39;t have a wayland driver back then. that is kind of a silly request.
02:55azakiit ran fine on intel and amd opensource drivers.
02:55ali1234raspberry pi doesn&#39;t have an x11 driver
02:56ali1234that didn&#39;t stop him
02:56ali1234it would be exactly as fair a comparison
02:56azakiyou wont see as much of a delta on desktop systems though. x11 has already been compared against wayland on embedded hardware that does have x drivers.
02:57ali1234whichis to say not fair at all
02:57azakiyou know that daniel stone worked on X11 for years right?
02:57azakilike well over a decade
02:57ali1234i know he worked on it for about 4 years, yes
02:57azakipretty sure it was more. =p
02:58azakihe&#39;s one of like three people that understands the X11 input stack.
02:58ali1234maybe if you count from his very first contribution to his most recent one
02:58ali1234he was only a maintainer for a couple of years
02:58ali1234also its ironic that he maintained the input subsystem, when in wayland, input is a SEP
02:58azakihe was a pretty huge contributor.. and paid to work on it by companies for years...
02:59azakihe knows X&#39;s limitations in embedded setups more than anyone, he worked for companies that were trying to ship embedded linux devices running X11 for years
03:00ali1234most of that time he was also a wayland developer
03:00azakino....
03:00azakihe was an X developer for many years before wayland sta<