OpenBSD Journal

bsdtalk # 106 - Interview with Matthieu Herrb about Xenocara

Contributed by dwc on from the bushy-nose-pleco dept.

The latest of Will Backman's bsdtalk series is online. Installment #106 is an interview with Matthieu Herrb about Xenocara, which is now in the tree!

File Info: 7MB, 14 Min [mp3 or ogg]

(Comments are closed)


Comments
  1. By Anonymous Coward (72.193.216.83) on

    You've got to be kidding?! Horrible interview. You can't make out anything that Matthieu is saying in this phone interview!

    Comments
    1. By Anonymous Coward (72.73.106.205) on

      > You've got to be kidding?! Horrible interview. You can't make out anything that Matthieu is saying in this phone interview!
      >

      I'm sorry that the audio was not very good, and some parts are hard to understand.

      Comments
      1. By Anonymous Coward (80.227.208.91) on

        > > You've got to be kidding?! Horrible interview. You can't make out anything that Matthieu is saying in this phone interview!
        > >
        >
        > I'm sorry that the audio was not very good, and some parts are hard to understand.

        BT = BSDTalk.

        BT: Welcome to the Show.

        MH Hi,

        BT: So could you start by telling us a little bit about yourself and what you do?

        MH: I'm living in France, in Toulouse in the South West of the Country and I've been working in France and Open Source projects and BSD for more than 10 years now. And I've been porting the X windows system to NetBSD first and now OpenBSD. I think this is why you are interviewing me today.

        BT: Yeh, and how did you first get interested in X Org and Unix in general?

        MH: I got interested in Unix when I was a student in an around Unix/Linux as a student in the lab where I prepared my PHD thesis I worked on a lot of Unix machines mostly [??] but also a few different machines so when in 1993 the first free Unix/Linux able to run a comp C (C compiler?) became available I became caught in that and since I [??] the people who work [386?] by this time I started looking at linux 386 and getting it to work on my hardware then I got involved and never stopped

        BT: I think Alot of people consider working on X to be a really big undertaking. What was you background in programming or graphical environements?

        ....would someone else like to decipher somemore?.....

        Comments
        1. By Bertrand Janin (tamentis) tamentis@neopulsar.org on http://neopulsar.org/

          > ....would someone else like to decipher somemore?.....

          Here is your previous deciphering + some more, anybody for the rest?

          BT: Today on BSDTalk, we're speaking with Mathieu Herrb, welcome to
          the show!

          MH: Hi!

          BT: So could you start by telling us a little bit about yourself and
          what you do?

          MH: I'm living in France, in Toulouse in the South West of the Country
          and I've been working in various Open Source projects and BSD for
          more than 10 years now. I'm involved in porting the X windows system
          to NetBSD first and now OpenBSD. I think this is why you are
          interviewing me today.

          BT: Yes, and how did you first get interested in X Org and Unix in
          general?

          MH: I got interested in Unix when I was a student in an around Unix as a
          student in the lab where I prepared my PHD thesis I worked on a lot
          of Unix machines mostly Sun and [???] but also a few different
          machines. So when in 1993 the first free Unixes able to run on our
          home PC became available, and I got caught in that. Since I had
          a graphic card that didn't work with xfree86 at this time, I started
          looking at xfree86 and getting it to work on my hardware, then I got
          involved and never stopped.

          BT: I think a lot of people consider working on X to be a really big
          undertaking. What was you background in programming or graphical
          environements ?

          MH: Well, I learnt these kind of things from scratch when I was first
          exposed to X, the Sun machines that we were running at my lab's by
          this time were [??] and it was even before the full IP connectivity,
          so one day we got one tape from someone coming back from the United
          States, "Hey I have the new graphical system that we will be using
          on our machines in the future, it's called the X Window System" it
          was X version 10 at this time. I had interest in these kind of
          things, so I took the tape and tried to install it on Sun 3 systems
          at the lab, took 20h to build and I started to explore X. Fifteen
          years ago X was a lot smaller than it is today, it was a lot easier
          to start with it.

          BT: Isn't the OpenBSD X perhaps a slightly different version of X than
          the standard X.Org?

          MH: No, not exactly. We have a few local changes to make X fits better
          OpenBSD before installation but basically it's X.org code, it's just
          a few configuration paramater switched for OpenBSD and a few local
          patches. Most of the patches we have in OpenBSD are not visible for
          users, so it doesn't really changes. What is different is that
          OpenBSD doesn't ship [??] and different environements, so for
          someone used to a modern Unix distribution which comes with [??]
          going back to fvwm as a windows manager can be strange, but this is
          not X.

          BT: And recently X.org has made some major changes to their
          architecture, I think it was version 7, can you describe what those
          changes are?

          MH: Yes, the big change is that they changed the way X is built, they
          switched from the imake based system which has been very specific to
          the X Window System to the GNU autotools. And instead of having one
          building script making the whole X Window System in one step, they
          decided to go for a modular architecture with one package per
          library, one package for the X server, one package for each drivers.
          So the total amount of new packages in the new X.org modular tree is
          above 300 that needs to be build separately, in the right order. For
          OpenBSD, since we want to be able to build X directly in the base
          system, we needed to do something to manage those 300 packages to
          build, and this is the goal of the new Xenocara project in OpenBSD.

          BT: So before Xenocara, to build X, you need to install some additional
          packages, like tcl or ticle. Is is still the case?

          Comments
          1. By Anonymous Coward (216.220.225.229) on

            MH: Yes. The big change is that they changed the way that X is built. They switched from the imake based build system which has been very specific to the X Window System to using the GNU autotools to build X. And for that, instead of having one just one build configure script that would build the whole X Window System in one step, they decided also to go to a modular architecture and several different modules, one package per library, ??, one package for the X server, one package for each drivers. So the total amount of new packages in the new X.org 7 modular tree is mare than 300 packages that have to be built separately, in the right order. For OpenBSD, since we want to be able to build X directly in the base system, we needed to do something to manage those 300 packages to build, and this is the goal of the new Xenocara project in OpenBSD.

            BT: So before Xenocara, to build X, I think I remember that you had to install some additional packages, like tcl or "tickle". Is that still the case?

            MH: No, that's no more the case. That was really a small side effect of just one utility on the Xfree86 3 packages, the xfree86setup program that needed Tk as a graphical user interface library. And since we wanted that program to be able to configure X, we needed Tk to build this particular application. In the next version of OpenBSD, 4.2, when we are effectively going to switch to Xenocara, we will probably lose those old xfree86 3 servers, so we don't need those configuration programs based on Tk any more.

            BT: And OpenBSD supports a variety of architechtures. Is Xenocara used across all of them? (6:57)

            Comments
            1. By Anonymous Coward (216.220.225.229) on

              BT: And OpenBSD supports a variety of architechtures. Is Xenocara used across all of them? (6:57)

              MH: Yes, that's the goal. For now, we still have some problems on the most slow architectures, mainly Vax and SH-4 Landisk systems that are not yet using shared libraries, because the new x.org modular tree more or less assumes that you have shared libraries there. So yeah, there are still a couple of problems to solve. But the goal for OpenBSD 4.2 is to have all the architectures using Xenocara.

              BT: Is the technology behind Xenocara something that is specific to OpenBSD, or is this something that could be useful to other BSDs?

              MH: It could be useable for other BSDs. It's mostly just BSD makefile with one generic set of rules that is used in every makefile. I had a short look at porting it to NetBSD, but for example, NetBSD changes are still in their system macro in their implementation of make which makes it a bit more complicated..??.. But anyway, this is just a makefile, and basically if someone wants to fix that, they can just take the macros from OpenBSD and copy them back to NetBSD under a different name so it doesn't clash with the modifications that they did in their own macros, and it should work.

              Comments
              1. By Anonymous Coward (216.220.225.229) on

                BT: And besides Xenocara, what other work are you doing with X?

                MH: One of my long-term plans is to work on X security. For now, I am still acting as the security officer of x.org. I am managing all the security related bug reports that come to x.org in general, but I also want to be able to work on fixing bugs more proactively in x.org code base. That's because of all the work involved in Xenocara. In the last year I have written a lot on this front.

                BT: Do you work with, or do you have any plans to work with other people who are trying to bring hardware acceleration to OpenBSD?

                MH: Yes, I would love to. The problem is that we don't have a lot of volunteers for now. I hope that one of the benefits of the modular build system in X is that we will be able to get more developers paying attention to X because its a bit easier now to start hacking on one particular part of X for instance, adding hardware acceleration for 3D.

                BT: And if people want to find out more information about Xenocara, where can they go?

                MH: They can go to the xenocara.org web page, which has a link to a talk that I gave to the last Fosdem in February. I have some slides that I made. They can always ask questions on the x11@openbsd.org mailing list, and I'll try to answer, or other people who know can also answer.

                BT: And where did you come up with the name for the project?

                MH: Well, it was a bit of luck. You know that we are using the fugu fish as the main OpenBSD logo, and I wanted to find a name which was the name of this fish that is cleaning the aquarium from the inside that you see in some aquariums. And I started looking on the web for this fish, and I found several different names, and one of the possible names was this latin name Xenocara which was fine because it also started with an X.

                BT: Alright. Well, are there any other topics that you want to talk about today? (11:06)

                MH: I think one of the most important things with the X Window system is that its one of the open source projects that was able to run on the biggest variety of different operating systems, from different Linux variants to Windows, to Mac OSX, to a lot of different OSes. And one of the difficult points of today's x.org direction is that they seem to care less and less about systems that are not Linux. This is a bit of a shame because not all the world is Linux, and this should not be the goal for any operating system, to be the only one system in the world. So this is one of the current difficulties in X. Its more and more difficult to get other developers of x.org to understand that the x.org project should take care of keeping its portability to a wide range of different systems and different hardware architectures.

                BT: I guess we can just encourage people on other operating systems to submit bugs upstream so that x.org is aware that there are many other users of different systems.

                MH: Exactly. I think, currently the number of bug reports is not representative of the importance of the different systems. Most people are just discouraging themselves from sending bugs if something doesn't work on that system because the consider it a minor system, I don't want to bother these guys with my problems. I think this is a wrong calculation.

                BT: Alright. Thank you very much for taking the time to speak with me today.

                MH: You're welcome.

                Comments
                1. By Anonymous Coward (65.243.149.146) on

                  Thanks for taking the time to transcribe all of that.

                2. By Dean (63.225.97.39) on

                  Thanks too! Much appreciated. Understanding/using X is a lifetime pursuit.

    2. By Anonymous Coward (213.41.176.184) on

      > You've got to be kidding?! Horrible interview. You can't make out anything that Matthieu is saying in this phone interview!
      >

      Yay X. always something new to hate. -- Adam Jackson

  2. By Anonymous Coward (216.220.225.229) on

    Full transcript here. Thank you to everyone who helped with this.

Latest Articles

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]