Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Operating Systems Linux

Interview With Fedora Project Leader Matthew Miller On 15 Years of Fedora (techrepublic.com) 48

Posted by msmash from the closer-look dept.
intensivevocoder writes: Fedora -- as a Linux distribution -- will celebrate the 15th anniversary of its first release in November, though its technical lineage is much older, as Fedora Core 1 was created following the discontinuation of Red Hat Linux 9 in favor of Red Hat Enterprise Linux (RHEL). Five years after the start of Fedora.next, the distribution is on the right track -- stability has improved, and work on minimizing hard dependencies in packages and containers, including more audio/video codecs by default, flicker-free boot, and lowering power consumption for notebooks, among other changes, have greatly improved the Fedora experience, while improvements in upstream projects such as GNOME and KDE have likewise improved the desktop experience. In a wide-ranging interview with TechRepublic, Fedora project leader Matthew Miller discussed lessons learned from the past, popular adoption and competing standards for software containers, potential changes coming to Fedora, as well as hot-button topics, including systemd.
This discussion has been archived. No new comments can be posted.

Interview With Fedora Project Leader Matthew Miller On 15 Years of Fedora More

Interview With Fedora Project Leader Matthew Miller On 15 Years of Fedora

Comments Filter:

  • Let me catch you up. (Score:3, Insightful)

    by Seven Spirals ( 4924941 ) on Wednesday October 16, 2019 @01:58PM (#59315266)
    15 years of Fedora? DBUS, XML, NetworkManager, GNOME3, PulseAudio, XDMCP-free Wayland, and the biggest loser of them all and the final stop on the Fuck Linux Forever Tour: Systemd/Journald. I've never been able to stand Fedora because of their support of the most M$Windowsy crap they can find. At least they are consistent, though. It doesn't matter though, all the clueful Linux users abandoned ship somewhere around 2015 (for places like FreeBSD and other greener pastures). Go to a technical conference and start spouting praise about Fedora then see if you can finish your sentence over the gails of laughter that follow.

    • Well put.

      I wonder who modded this a troll?

    • Re: (Score:3, Interesting)

      by Dadoo ( 899435 )

      I don't know about anyone else, but the systemd bashing on this site is starting to get a little old. I started using Unix in 1985, and having heard about how it was designed with multitasking in mind, I was pretty excited to learn as much as I could about it. Then, I learned how the system startup process worked, and I couldn't believe it. So much for multitasking.

      The truth is Unix (and Linux) should have had a parallel startup process back then, and I for one, am happy we finally have one. Is systemd perf

      • Unix (and Linux) should have had a parallel startup process back then

        Upstart [debian.org] maybe?

      • Re: (Score:3, Informative)

        by Etcetera ( 14711 )

        I don't know about anyone else, but the systemd bashing on this site is starting to get a little old. I started using Unix in 1985, and having heard about how it was designed with multitasking in mind, I was pretty excited to learn as much as I could about it. Then, I learned how the system startup process worked, and I couldn't believe it. So much for multitasking.

        Unix has a multitasking model; it's called forking. Something which can be fired off to complete at an arbitrary time can be forked and forgott

        • "Unix has a multitasking model; it's called forking."

          I read this and immediately knew we were all in for a post full of misinformation from someone with no clue about how Linux works and it wasn't hard to guess you would be clueless about systemd as well.

          "You're conflating service startup and boot."

          No. You did that. He was clearly talking about system startup, aka system boot. We know this because he was talking about the init system, to wit, systemd. We also know this because he rightly asserted that SysV

          • I read this and immediately knew we were all in for a post full of misinformation

            I'll agree the statement is a bit clumsy but he's still fundamentally correct. Unix is, in fact, multitasking. It does, in fact, use fork() (as well as threads and other more fundamental mechanisms) to achieve that. So, what? We are just supposed to believe that your emotional reaction and insults to the guys post somehow makes *technical* sense to people looking at your content-free response?

            from someone with no clue about how Linux works

            He made and accurate statement and your denial only makes you look like the clueless one.

            and it wasn't hard to guess you would be clueless about systemd as well.

            Well, I'm your hucklebe

            • "Unix has a multitasking model; it's called forking."

              "I'll agree the statement is a bit clumsy but he's still fundamentally correct. Unix is, in fact, multitasking. It does, in fact, use fork() (as well as threads and other more fundamental mechanisms) to achieve that."

              Multitasking is not called forking moron.

              "It's PID #1 - just like init was!"

              This speaks to your fundamental lack of knowledge of systemd. Yes there is an init process with PID 1 as on ALL Linux systems. Yes it is the init process:
              PID 1 Owne

              • Wow, so you are doubling down that systemd isn't PID #1? You are just too stupid to live, sometimes, man. How do you even remember to breath. I'm sitting there showing you a stock RHEL7 box with a systemd process clearly listed as PID 1 with PPID 0 and you are so fucked up that you can somehow ignore that, simply show from some system that it might have more than one process that's not PID 1 (which I already said, you fucking retard). Yeah, Systemd is multi-process. We already talked about that, remember du

                • "Wow, so you are doubling down that systemd isn't PID #1?

                  For those who actually want to know what they are talking about you can read this [stackexchange.com], and this [0pointer.de]. We know this idiot won't do it though because he revels in his own ignorance.

                  • Dude face it: you are fucking wrong. Systemd is PID #1 and that's just a fucking fact. Everywhere. All the time. The fact that's not monolithic is not what I said. It's what you implied when you could no longer argue. I never once said it because I actually work with Systemd enough to know what it is and what it isn't, while you, on the other hand just talk shit out of your ass about things you've never seen based on idiotic blog entries like the one you are linking here where least half the "bullet points"

      • Re: (Score:2)

        by mattdm ( 1931 )

        Yeah, pretty much my take too.

        Thanks for attempting to engage in a constructive dialog. I came here to see if there were any useful comments, and, wow, all of the jokes about Slashdot going downhill from a decade ago are calling and want the future back.

        And what's with these 8-digit user IDs? *shakes cane at sky*

      • Why SystemD won (Score:4, Informative)

        by steveha ( 103154 ) on Wednesday October 16, 2019 @06:16PM (#59316328) Homepage

        I'm sorry, though, SysVinit is amateur hour

        I wouldn't go that far. It was a good solution when it was first written. But I do feel the time is already past when it was no longer the best solution.

        Anyone know why we're not using [Upstart], instead?

        Because the people who actually have to do the work of packaging Linux have collectively decided to use SystemD rather than Upstart.

        I, myself, do not have the knowledge or experience to make this judgement call. But I have read things written by the people who did make that judgement call. I was very interested when the Debian Technical Committee decided that SystemD was the best choice, and this very thoughtful commentary really explains why. This commentary also specifically discusses why Upstart was not chosen instead of SystemD. (In fact, there were three contenders: OpenRC was the other one that lost out to SystemD.)

        https://lists.debian.org/debian-ctte/2013/12/msg00234.html [debian.org]

        Thoughtful analysis like this posting won me over to tentative acceptance of SystemD. I have not seen anything as thoughtful from the anti-SystemD discussions, which I can boil down to a few points:

        0) I just hate SystemD!
        0a) SystemD is Not The UNIX Way!

        Included in these points are the idea that SystemD is "monolithic". As far as I can tell, SystemD is composed of a bunch of reasonably small pieces that each do one thing, which is more or less the UNIX Way. However, a criticism of SystemD is that these small pieces were all designed by a small cabal of people and all designed to work together, and taken collectively add up to a monolithic design. I can imagine a sufficiently dense set of mutual dependencies effectively welding a bunch of small tools into a monolith; but I haven't read a thoughtful analysis making that point, just emotionally-charged claims.

        In principle, it should be possible to swap out pieces of SystemD, or implement SystemD interfaces on BSD. In practice this is somewhat difficult because the pieces of SystemD are not well documented; it's an ambitious project that's moving fast, so anyone wanting to re-implement pieces will have to first figure out what all the features and requirements of the pieces are before it can happen.

        Note: I read a Slashdot discussion about how KDE is getting rid of some of its code and depending on SystemD, and the KDE guys said that KDE would still run on BSD because someone was implementing those pieces for BSD. Here's a link to my comment from that discussion thread:

        https://slashdot.org/comments.pl?sid=8387451&cid=51004177 [slashdot.org]

        1) I tried SystemD and Bad Things Happened!

        These complaints are difficult for me to evaluate. I have simple needs and SystemD is working very reliably for me, so I'm not seeing any Bad Things. Also, I haven't seen anyone do the homework to show that any particular Bad Things represent an inherent design flaw in SystemD. Any complex software will have bugs. Bugs can be fixed while bad design is a problem forever.

        2) The SystemD people are Not Nice People!

        That may be true and it may not be, but I'm more interested in technical merit.

        Also, vocal members of the Linux community started in with this abuse before SystemD was even introduced, as its main architect was also the architect of PulseAudio.[0] So, how much of any bad attitude on the part of the SystemD team reflects a bad reaction to abuse hurled their way? "Lennart is Literally Hitler. Also, I didn't like the tone of his reply after I explained to him why he's such an idiot and his designs are all garbage, so he's a jerk as well as incompetent."

        P.S. When I was looking for the Debian discussion link I gave, above, I found this Reddit page with links to other analyses, both pro-SystemD and pro-Upstart.

        • Did you know OpenBSD starts up with a single script (/etc/rc)? Are those guys who write OpenBSD and OpenSSH a bunch of morons for not adopting Systemd or it's brethren ? See they know that a real Unix guru doesn't solve problems by heaping on complexity, quite the opposite. Love how you made up all your own points to argue with and completely ignored the shit security history [mitre.org] of Systemd since that is a little to hard to straw-man and hand wave off (along with the chicken & egg issues with Journald's b

          • Re: (Score:3)

            by steveha ( 103154 )

            Are those guys who write OpenBSD and OpenSSH a bunch of morons for not adopting Systemd or it's brethren ?

            Nope. Nor are the Devuan [devuan.org] guys a bunch of morons, as far as I know.

            I think it's great to have multiple, competing solutions to the same problems. Also, I have great respect for people who actually do the work to maintain BSD or a Linux distribution.

            In my turn, I'd like to ask you: are the guys who maintain Debian [debian.org] a bunch of morons? How about the guys who maintain Fedora, OpenSuSE, Arch and Mageia? A

            • Nope. Nor are the Devuan [devuan.org] guys a bunch of morons, as far as I know.

              Being a part-time Devuan user, let me answer that: no. It rocks about as hard as Linux is able to these days. It still has the scent of Debian when it was cool.

              I think it's great to have multiple, competing solutions to the same problems.

              I agree somewhat, but I don't think that means I'm going to endorse all those "solutions" either. A lobotomy is a solution to mental illness and systemd-fanboyism, but I'm not advocating for it.

              Also, I have great respect for people who actually do the work to maintain BSD or a Linux distribution.

              Thousands work to help maintain distributions. I agree their stock goes up for that. However, if they want to come shitpost for Systemd on /. it doesn't mean

    • Re: (Score:3, Insightful)

      by slack_justyb ( 862874 )

      15 years of Fedora? DBUS, XML, NetworkManager, GNOME3, PulseAudio, XDMCP-free Wayland, and the biggest loser of them all and the final stop on the Fuck Linux Forever Tour: Systemd/Journald.

      Okay let's address those then.

      DBUS

      Well let's look at what it replaced. CORBA [wikipedia.org] and DCOP [wikipedia.org]. Now to the end user, both of these models offer a very nice easy to script interface.

      dcop $SOME_ID konsole sessionCount

      Would get you the number of terminals open. Compare that to VLC controlling volume via DBus.

      qdbus org.mpris.MediaPlayer2.vlc /org/mpris/MediaPlayer2 org.freedesktop.DBus.Properties.Set org.mpris.MediaPlayer2.Player Volume 0.5

      And this isn't me cherry picking here. A lot of CORBA and DCOP was very easy

      • Re: (Score:2)

        by deKernel ( 65640 )

        Sir,

        May I buy you a virtual beer...and not one of those crappy beers, but a good beer?

      • That was an excellent post. I literally don't remember ever feeling like a post was so good that I wanted to reply in this fashion. Well done. It will be interesting to see how many people can't understand it and are foolish enough to even try to counter your points, which should be studied by everyone who has ever made the mistake of reading these "Linux / systemd / gnome / wayland / etc. sucks" trolls and believing the person saying it has any idea what they are talking about.
        • It was an uninformed shit-post by a systemd fanboy. I'm not surprised you liked it. You've always been a fucking idiot on /. as long as I've seen your nym.
          • I wrote:

            "It will be interesting to see how many people can't understand it and are foolish enough to even try to counter your points"

            I stand corrected. It turns out only one of the most boring and infamously clueless Slashdotters was stupid enough to try it.

      • Okay let's address those then.

        A for effort and verbosity, F for you still sound like a minor league Fedora dev.

        DBUS Well let's look at what it replaced.

        Nice attempt at a straw man. You make the assumption that the whole idea is even desirable is the problem. It's the whole approach to inter-process communication that's the issue, not the previous crappy idea in the same vein. Not all desktops and/or Window Managers even use that shit and praise $diety that they don't make the same assumptions you do that we are all going to need CORBRA-like functionality done with some GUI-f

        • Nice attempt at a straw man. You make the assumption that the whole idea is even desirable is the problem. It's the whole approach to inter-process communication that's the issue, not the previous crappy idea in the same vein. Not all desktops and/or Window Managers even use that shit and praise $diety that they don't make the same assumptions you do that we are all going to need CORBRA-like functionality done with some GUI-friendly API, because some of them still don't suck like *box and the plethora of tiliing window managers? Oh, I didn't include a Fischer-price desktop mountain-of-packages-pile-of-crap to attack and show how dependent it should all be on bad IPC? Right. That's because I neither use it or advocate for that kind of tripe. I'm for keeping a Unix a developer friendly OS, not kowtowing to mindless touchscreen tablet users ala Fedora's latest flavor-of-the-month directions.

          What are you even going on here about? Are you indicating that IPC isn't needed or are you indicating that IPC should be handled some other way that you failed to mention? The entire point here is that processes need to talk to each other. Pipe and UNIX sockets provide one way of doing that, but it is incredibly limiting in how to go about serializing objects across those methods. Simple message passing between services are fine with it, but wanting to have something more than "Okay I'm ready!" go acros

          • What are you even going on here about?

            I'll simplify my point, because this isn't something I give a major damn about. I think the idea of an ORB is usually wrongheaded in the weeds/details of how a request broker works and often implemented extremely poorly. I prefer other methods of IPC (like *any* other). I didn't say IPC sucks, I put DBUS in a shit-list.

            Okay look, I'm not some Fedora fan boy here.

            It's what you started off sounding like to me. The extra nuance helps. Just remember that when it looks like a duck...

            What I am saying is that developers are going this route

            I agree. When lemmings run over a cliff I'm not going to stop them.

Slashdot Top Deals

On the eighth day, God created FORTRAN.

Close