Announcement

Collapse
No announcement yet.

Will Wayland Become A New Desktop Standard?

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    I'd do this to have only one IPC protocol, which already has nice features for access control.
    BEEP looks nice, but as far as I understand it doesn't specify marshaling formats.
    XML, XDL and (IDL - Vortex will detects and coverts IDL to XDL) marshalling all data, and unmarshall all replies received.


    If your system is available over the network then it can be rooted (even if in theory).
    Sure in theory. AES-256 can be cracked in theory doesn't mean we will ever see it happen in our life time. And AES isn't
    even the strongest there are others that are stronger and in the public domain.
    Last edited by zester; 06-12-2011, 03:26 PM.

    Comment


    • #32
      Originally posted by zester View Post
      XML, XDL and (IDL - Vortex will detects and coverts IDL to XDL) marshalling all data, and unmarshall all replies received.
      Sorry, that disqualifies it then. XML is too slow for fast IPC (it's fine for service descriptions, but too heavy for IPC payloads).

      Sure in theory. AES-256 can be cracked in theory doesn't mean we will ever see it happen in our life time.
      There were remote exploits in the TCP stack. Are you sure that there aren't any still present? Or had you turned off TCP as well?

      Comment


      • #33
        I'd do this to have only one IPC protocol, which already has nice features for access control.
        BEEP looks nice, but as far as I understand it doesn't specify marshaling formats.
        XML, XDL and (IDL - Vortex will detects and coverts IDL to XDL) marshalling all data, and unmarshall all replies received.


        If your system is available over the network then it can be rooted (even if in theory).
        Sure in theory. Any type of security measure can be broken, but that doesn't mean it has or is common like the OP made it out to be.

        Comment


        • #34
          Maybe I should have refrazed those responses.

          1. You wouldn't want an IPC framework in Wayland because Wayland was mean't to be as abstract as
          possible. So that it will scale properly in the future. And you wouldn't wan't to tie all toolkits to a dbus
          dependence. They might wan't to implement there own IPC framework.

          2. Rootkit's on Linux are possible but not all that common. You see them mostly on servers that are using outdated software.

          The Linux kernel actually has mechanisms to protect against rootkit's.
          And if a Linux user implements proper security (Read only Partitions, ACL, Real-time Disk Encryption, Strong Passwords, Proper Network Port Management ,Keeping your System Updated with the latest Security Fixes) Then yes chances that your Linux box will be rooted are unlikely.

          Note: That is considered entry level security there are much more advanced bare-metal ways to protect a Linux box but there considered out of scope for your average Linux user.
          Last edited by zester; 06-12-2011, 04:49 PM.

          Comment


          • #35
            Sorry, that disqualifies it then. XML is too slow for fast IPC (it's fine for service descriptions, but too heavy for IPC payloads).
            DBUS uses XML all over the place. Besides Vortex uses libaxl for XML it's 14% faster that libxml

            It has XML-IPC (An inter-process communication protocol built upon file pipes and
            the methodCall and methodResponse encodings of XML-RPC.)

            Not to mention non-blocking parallel communication

            And for the IDL to XDL it uses a protocol compiler.

            Stop acting like you know what your talking about.
            Last edited by zester; 06-12-2011, 05:06 PM.

            Comment


            • #36
              Originally posted by zester View Post
              And if a Linux user implements proper security (Read only Partitions, ACL, Real-time Disk Encryption, Strong Passwords, Proper Network Port Management ,Keeping your System Updated with the latest Security Fixes) Then yes chances that your Linux box will be rooted are unlikely.
              Same could be said about all the mainstream OS's, a system is only as secure as the competency of the person hardening the system.

              Comment


              • #37
                Originally posted by deanjo View Post
                Same could be said about all the mainstream OS's, a system is only as secure as the competency of the person hardening the system.
                Not exactly. Windows is insecure by design. It doesn't even ask for password in most (any?) cases.

                Comment


                • #38
                  I thought this thread was about Wayland.

                  For me, it doesn't really matter if the X server is re-written or replaced, as long as it works without problems. The developers have decided to make a new display server instead of patching the almost 25 year old X. I think they've made a good decision, they know the inside of X better than the average forum member who just uses X and does not worry about what happens inside.

                  X is becoming a mess of old and new code. There are for example 3+ ways to get dual-screen which aren't compatible with each other.

                  Comment


                  • #39
                    No, I'd rather they overhaul the X standard into X12 than see something as immature as Wayland take over.

                    Wayland makes the mistake of requiring KMS instead of merely supporting it as a bonus, assuring none of the best drivers Linux has will support it all that well, if at all. KMS and OpenGL aren't mutually exclusive, but you'll be hard put to find any driver with complete KMS and accelerated OpenGL support.

                    I'm being tactful. Here's me being blunt: Such drivers do not yet exist.

                    Sure, there's some somewhat DECENT support for accelerated OpenGL in a good deal of the open source drivers. Complete? No. Enough to fully support, say, KWin's compositing? No.

                    I like the ideas of Wayland wants. Pixel perfect frames every time sounds very idealistic, but likely hard to achieve with good performance with the state of the current drivers available for Linux that still support KMS. If they simply had KMS support optional like X did we'd actually have something to talk about.

                    But, in all honesty, why reinvent the wheel when you can improve the wheel we have and save some effort and keep some compatibility? Let's not switch to Wayland, instead, lets take what we learned from it and implement it in a new X standard: X12.

                    In X12 we'd rid ourselves of the cruft that started building up over the all-too-long run of X11, add in a lot of the concepts and design of Wayland, while still allowing things like low-level hardware support and network transparency to work. Ultimately, X12 should do what X11 doesn't in addressing the needs and designs of the modern desktop.

                    Comment


                    • #40
                      What makes you think that Wayland is not X11 with accumulated experience?

                      Server-side rendering is an evolutionary dead end. It's needlessly complex and not needed at all for modern mobile devices. So Wayland is the future.

                      Comment


                      • #41
                        Originally posted by Cyberax View Post
                        Server-side rendering is an evolutionary dead end. It's needlessly complex and not needed at all for modern mobile devices.
                        It is not dead (see Nintendo's latest Wii U console) but the applications have changed.

                        Comment


                        • #42
                          Many people here are calling Wayland 'immature' - for the current state of Wayland, that may indeed be true. But it will still take a couple of years of development before any mainline distro will start to use Wayland. I am sure that Wayland will be greatly improved in the following years.

                          Comment


                          • #43
                            Originally posted by Yaro View Post
                            No, I'd rather they overhaul the X standard into X12 than see something as immature as Wayland take over.
                            Sure, there's some somewhat DECENT support for accelerated OpenGL in a good deal of the open source drivers. Complete? No. Enough to fully support, say, KWin's compositing? No.
                            Ahem, the only drivers supporting the new kwin approach by using OpenGL ES is the oss drivers. Neither of the proprietary drivers support OpenGL ES properly in order to run kwin. So your above statement is wrong. I just tried kwin with r600g with kwin using OpenGL ES and the performance is incredible fast (this is with stable mesa 7.11). It is much faster than Catalyst with/without OpenGL 2.0 shaders.

                            AFAIK if you are using nouveau, you have to use mesa GIT.

                            Comment


                            • #44
                              Originally posted by chithanh View Post
                              It is not dead (see Nintendo's latest Wii U console) but the applications have changed.
                              As far as examples go, that's kind of a different case. The Wii U pad thingy is not a separate device with its own OS and application stack and environment. It's really just another LCD that's attached to the console, except it's 'attached' wirelessly instead of via a display cable.

                              Now, if the Wii U pad had its own small OS and its own GPU and the Wii U was just sending rendering commands to the pad, then it would be a good example of server-side rendering in the context of X11.

                              Originally posted by zester
                              DBUS uses XML all over the place. Besides Vortex uses libaxl for XML it's 14% faster that libxml
                              D-Bus does not use XML for its actual IPC, at all. It allows service descriptions to be written in XML, which can be queried _if_ an application wants it, but that's really only good for debugging. You don't query a service description at runtime because getting a random list of methods does absolutely nothing usful for the software: all it cares about is whether the other end implements a specific well-defined previously agreed upon interface or not, which is not queried via XML. When D-Bus does serve that XML, it is basically just blitting it into the network buffer, and not processing it. D-Bus also uses XML for its configuration and automatic startup config, but these are read once upon startup (or when changed) and no XML processing is going on during message transmission. The actual messages sent over DBUS are a fairly efficient binary format.

                              Originally posted by kraftman
                              Not exactly. Windows is insecure by design. It doesn't even ask for password in most (any?) cases.
                              This is wrong on several levels. First, Windows _is_ just as secure as its user. It's quite possible to have run Windows XP for years with no anti-virus and never haven gotten a virus. The trick is to not download and run OMG_CELEB_BOOBIES.EXE. It's a subtle, difficult thing to master. The same kinds of users who would fail to realize the inherent danger in such an activity, however, are going to be just as stupid in Linux, and when they see a page that says "OMG CELEB BOOBIES, DOWNLOAD THIS AND TYPE SUDO INSTALL_BOOBIES.SH TO INSTALL BOOBIES ON YOUR DESKTOP" they will gladly do that very thing and think nothing of entering their password ("why is linux so lame and asking for passwords all the time? so dumb. just annoying like windows 7 and its uac prompts always asking me if im sure if I want boobies on my desktop! and why is firefox so slow lately anyway?!").

                              Windows Vista/7 especially are quite secure. I will steadfastly hold to the claim that they are in fact more secure desktop OSes than Linux+X11 by a long shot. That is due to a significant effort put into very fine-grained privilege separation in services, the UAC dialogs and the security of how those are implemented (where's the "please make sure this full-screen window that may or may not be my screensaver unlock dialog really is my screensaver and not something else?" for X11 at?), the significantly more thoroughly protected file system, etc. Sure, it won't ask for a password, but if you actually think about this with any level of critical cognition, why would you care about a password prompt rather than UAC? Nobody but the physical user sitting in front of the machine can agree to a UAC prompt, while on Linux/X11 pretty much anything and everything can snoop on half your password dialogs (did you just put in your SSH key in that root-owned dialog? who cares, any app can snoop on the X11 events to that window trivially, because nobody is actually using XACE for anything useful yet) and respond to password dialogs on their own. Honestly asking a user to type a password does not protect him from malware any more than clicking a properly protected button does, because either one requires the user to consciously decide to do something stupid. The whole purpose of such things is to verify the human's intent, and the Windows implementation actually succeeds at that while the Linux implementations do not.

                              Password prompts are built around the assumption that someone else is physically at your computer trying to mess it up; let me tell you, if someone is physically at my desktop, they've already broken into my house and I have bigger problems than my PC; and if someone got into my laptop, that means they already got past one password prompt at the login/unlock screen and clearly a second password prompt is not going to deter them further. Linux is just obnoxious when it even tries to be secure, while Windows is effective and as unobtrusive as possible.

                              The Linux desktop has things to learn and emulate from the modern Windows experience. Laughing at something you obviously haven't used or understood while trumping up the out-moded and poorly-conceived UNIX tradition of password prompts is not helping you, Linux, or anyone else to be more secure and safe.

                              Comment


                              • #45
                                not related to wayland but after reading elanthis post i stumbled uppon this:

                                http://venturebeat.com/2011/08/05/wt...rity-approach/

                                this comes from a guy that was hired to tackle the security problems in Win.

                                Comment

                                Working...
                                X