Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Wednesday November 19 2014, @11:25AM   Printer-friendly
from the I-hope-we-don't-regret-this dept.

Ian Jackson's general resolution to prevent init system coupling has failed to pass, the majority vote deciding that the resolution is unnecessary. This means that not only will Debian's default init be systemd, but packages will not be required to support other init systems. Presumably, this means that using other init systems on Debian (without using systemd as a base) will not be possible without major workarounds, or possibly at all. It also leaves the future of Debian projects such as kFreeBSD unclear, as systemd is linux specific.

The vote results can be found here

The winners are:

Option 4 "General Resolution is not required"

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 1, Insightful) by Anonymous Coward on Wednesday November 19 2014, @06:57PM

    by Anonymous Coward on Wednesday November 19 2014, @06:57PM (#117803)

    Dont take this the wrong way... It is just something I have wanted to rant about for awhile

    Modularity works really well on a CLI when commands can be chained into a long pipe.

    I hate junk like that. It is modular but fairly unreadable gibberish usually. Just today I spent 30-40 mins decoding some crazy command line someone sent me. It is not always clear what particular parameters do. There is little continuity between programs so you end up digging out docs for every little thing. For example in one program 's' may mean recurse sub-directories and 'r' means turn the text red. Then in the next program 'r' means recurse and s means something totally different unless you use capital S.

    I can hang with the best CLI wizard out there but it really seems like command line pipe is a really badly abused interface. I look at the way it is used and think there *HAS* to be a better way. But I am drawing a blank (that may be a good thing :) ). GUIs usually make things a bad click fest. Input files/memory is 'hidden'. Scripts sorta of do it but then you have to think in context like a programmer (which I can do as well). So I grumble and use it thinking there must be something better.

    I am just not sure. It seems 'wrong' in some fundamental way that I can not put my finger on. Stuff like systemd I can see *why* they did it. Scripts are brittle in subtle ways. But in essence they are super simple. Yet the chaining of events creates a giant storm of bad in/bad out sometimes. The idea is a good one. But the method they went about to foist it on everyone is bad (simply ignoring what everyone was saying). Instead of taking what was there and extending it (like the system was designed) they threw it all out and basically said '20 years of scripting is garbage'. It is the same reason everyone abandoned xfree86 when it was convenient. There was an opportunity here to really make linux way better. Instead it is a bunch of infighting and a 50/50 split if you care or dont. It will be 10 years before whatever systemd ends up being, that it finally calms down and people undo the damage and creates a decent consistent startup/service interface out of it. Better consistency is one thing linux could do much better with.

    Thanks for the time letting me rant.. I am off to debug a startup issue on an embedded linux system.... Hopefully it is not a hardware issue (which means deracking 20 hds and weeks of RMA hassle from 8 states away).

    Starting Score:    0  points
    Moderation   +1  
       Insightful=1, Total=1
    Extra 'Insightful' Modifier   0  

    Total Score:   1  
  • (Score: 2) by Nerdfest on Wednesday November 19 2014, @07:13PM

    by Nerdfest (80) on Wednesday November 19 2014, @07:13PM (#117806)

    What you're complaining about is more an API problem than an architecture problem though. You can have a good architecture and a bad API (as you describe with some command line parms). I'd prefer to have that than the opposite. It's hard to fix a bad architecture.

    • (Score: 2) by Arik on Thursday November 20 2014, @12:51AM

      by Arik (4543) on Thursday November 20 2014, @12:51AM (#117925) Journal
      Too bad I dont have mod points, but that was a very good post. Concise, to the point, and very insightful.

      This is what the systemd proponents just cant seem to wrap their heads around. A bad design is a bad design, no matter how beautifully executed.
      --
      If laughter is the best medicine, who are the best doctors?
    • (Score: 0) by Anonymous Coward on Thursday November 20 2014, @03:29PM

      by Anonymous Coward on Thursday November 20 2014, @03:29PM (#118131)

      You can have a good architecture and a bad API

      I see what you are saying. But that still just does not seem right. Jamming things thru the CLI or thru stdin just seem wrong to me. They are very useful tools but it seems fraught and easy to create poor API.

      I am a big believer of datastructs first. If you design your data properly the API usually comes out of it quite cleanly. To your architecture point. Yet the way data is moved around seems clunky to me for some reason. I am not saying it is bad. It just seems weird and not natural.

      I cant put my finger on it. To use a quote from the matrix 'it is like a splinter in my mind'. It is like we are using the wrong interface for data movement. We are overloading meaning which creates confusion. Until something better comes along I will continue to use the crap out of it :)

      It's hard to fix a bad architecture.
      I agree... Which is why I am calling systemd a missed opportunity. It was an opportunity to remove one thing I see as very clunky in the 'distro wars'. The startup process. Each distro basically had their own versions of these scripts. Lots of redundant work just for the sake of being 'different'. Not necessarily better/worse, just different. It is one of the reasons many blew up on the idea. As it is part of the identity of the distro they picked. The systemd guys however cant seem to go back and fix what they have and are playing with polishing the brass on the rails of the titanic. They have not created a consistent interface, they 'grew one' then added in some weeds. If they get their interface right the other parts would fall naturally out of it and they would not have tons of work ahead of them. They could then go on and fix/create other things instead of farting around with how the system boots up. Which frankly is a minor part of my computer usage (at least it better be).

  • (Score: 2) by cafebabe on Wednesday November 19 2014, @07:53PM

    by cafebabe (894) on Wednesday November 19 2014, @07:53PM (#117824) Journal

    For example in one program 's' may mean recurse sub-directories and 'r' means turn the text red. Then in the next program 'r' means recurse and s means something totally different unless you use capital S.

    That irritates me too. I wish that -R and only -R was recurse. I think the most annoying examples are within the suite of aircrack-ng utilities. Even the second generation of a bunch of commands released together don't use consistent parameter conventions. However, if I make the comparison with GUI software, menu shortcuts (such as Alt-H for Help) are equally inconsistent.

    Stuff like systemd I can see *why* they did it.

    I assume your background is weighted towards Windows or MacOS. All I see is technical debt [soylentnews.org], Not Invented Here [soylentnews.org], "WTF is RFC5452? [soylentnews.org]" and "Do you hate [google.com] disabled people? [youtube.com]"

    Yet the chaining of events creates a giant storm of bad in/bad out sometimes.

    Computer pioneers usually had a mathematical background, so stuff like y=f(g(h(x))) was bandied around without having to consider implementation details such as exception processing. However, when foo | bar | baz is stateful or starts spewing garbage, it falls outside of this mathematical paradise.

    It will be 10 years before whatever systemd ends up being, that it finally calms down and people undo the damage and creates a decent consistent startup/service interface out of it.

    We just cannot determine what it will be yet and that alone makes it unsuitable for deployment on stable systems. Regardless, deployments will be cracked. Repeatedly. This isn't an academic issue or an opportunity for schadenfreude. Some of those systems may contain my private information and your private information.

    --
    1702845791×2