Forgot your password?
typodupeerror
This discussion has been archived. No new comments can be posted.

Evolution Of Debian Package Dependencies Resemble Predator-Prey Relationships

Comments Filter:
  • by m4c north (816240) on Wednesday November 23, 2011 @04:17PM (#38151776)
    FTFA:

    The team went back to 1993 and compiled statistics on every major stable release

    In other words, "The team compiled statistics on the last three major releases".

  • Pure nonsense (Score:4, Insightful)

    by Chemisor (97276) on Wednesday November 23, 2011 @04:21PM (#38151808)

    Once again, people versed in one discipline apply their skills to another with results that sound fancy and expensive, while really are just nonsense [abstrusegoose.com].

    • Re:Pure nonsense (Score:5, Informative)

      by Kamiza Ikioi (893310) on Wednesday November 23, 2011 @04:26PM (#38151856) Homepage

      Once again, people versed in one discipline apply their skills to another with results that sound fancy and expensive, while really are just nonsense [abstrusegoose.com].

      Not really. Trial and error is at the heart of evolution which can apply just as easily to living organisms, memes, or software (which is actually a form of meme). It just happens to work quicker on the latter. Predator-Prey relationships are merely an extension of evolution, and is all about inter-dependance. If a package depends on another, one could be considered predator, and the other prey. If development slows on that which is depended on, then the predator must find new prey or face lower numbers and/or extinction.

      The greater mistake is a myopic view of disciplines, thinking that nobody else can contribute to the understanding of another discipline except those who specialize in it.

      • by theghost (156240)

        But this doesn't seem to shed any new light on things. Does it give us new insight into what we should be doing or avoiding? Does it provide us with any new ways to deal with problems? Nothing is obvious to me off the bat, but time will tell if some novel application of this observation will bear fruit. Right now, it's just an interesting phenomenon.

      • by Synerg1y (2169962)

        Let me see if I can put this on context for you: Life doesn't debug, debian does. All flaws and successes in debian, the team behind it is responsible for. Thus a newer module does not "eat" a weaker module, the debian team gets spammed w complaints on how much a feature sucks or how a bug is costing stability and then fix it in turn making the module better. *shrug* article is very vague as well.

        • Re:Pure nonsense (Score:5, Insightful)

          by Kamiza Ikioi (893310) on Wednesday November 23, 2011 @04:42PM (#38152056) Homepage

          Let me see if I can put this on context for you: Life doesn't debug, debian does.

          It's called reproduction. Failure in life results in decreased chance of reproduction, including the ultimate decreased chance, death.

          Dependencies compete for limit resources called developers. Just look at all the various sound systems Linux has gone through. They, and the software that depends on them fight to gain more users to thus draw more developers. And, they either live and reproduce, or die.

          You are thinking of predator-prey wrong. It's not about eating by about parasitic relationships. And biology is very good at that. What this can show a developer is, based on known algorithms of life and chances of survival, which pieces of software are most in danger of extinction. And that's not always obvious. A music app might be facing extinction because of a particular reliance on a codec dependency that is slowly losing support in favor of another. It let's the developer see further down the road.

        • by izomiac (815208)
          Each successive generation of a species contains fewer "bugs" than its predecessor, as disadvantageous traits are selected against. This is the sole responsibility of mate selection by individuals. In software, a new module generally contains the useful feature set of the older one, so it's not whole inaccurate to say the newer module consumed the older one. User complaints are one of the major selective pressures that drive these processes.

          Ecology is an emergent property of evolution, which is an eme
        • "Life doesn't debug" - missed the Darwin memo, eh?
      • by pclminion (145572) on Wednesday November 23, 2011 @05:59PM (#38152976)

        If a package depends on another, one could be considered predator, and the other prey. If development slows on that which is depended on, then the predator must find new prey or face lower numbers and/or extinction.

        Dammit, this other package my package depends on is just too fucking stable! It never changes! Every day it's just the same damn thing again! I need to base my package on something that's more of a moving target. Otherwise I run the risk of my own software becoming... (shudder) STABLE.

        • by hawk (1151)

          In 96 or 97, this was solved when a major package--live or some such, on which nearly everything actually depended to execut--was declared by one person to have an impure license, andhe unilaterally pulled it. If you ran an update, you ended up with a system that w dead in the water, you could run a shell, but I don't remember whether it was single user or console.

          In the meantime, I tried FreeBSD afain, and that time it supported my hardware, and stayed with it (and discovered how FreeBSD & Linux would

  • by migla (1099771) on Wednesday November 23, 2011 @04:25PM (#38151852)

    I read the first linked FA. It doesn't explain, but just states that dependencies are a predator-prey relationship. Would that mean as in cats depend on mice (in a hypothetical ekosystem with not much more than mice and cats) and without no mice there would be no cats? If that's what they mean, then that is likely just a statement of the nature of packages using a biology analogy, not any kind of findings from their research.

    Was there something to take to heart about the things about modularity and conflicts and stuff? Too tired or ignorant to get that.

  • by Urban Garlic (447282) on Wednesday November 23, 2011 @04:26PM (#38151858)

    ... is here [xkcd.com].

    • This is funny just on the superficial level of a stick man being eaten by locusts, but is there some satirical element to it that i'm not getting?
  • After countless stints in Dependency Hell, I would ask: which is the predator, and which is the prey?

    • by OneAhead (1495535)
      Easy: Marc Ewing is the predator, you're the prey.


      What? I should have typed "Ian Murdock"? No no no, Debian doesn't have dependency issues.
  • Just because a mathematical tool works in other fields does not mean they are related. You can describe many things with differential equations.

  • by sconeu (64226) on Wednesday November 23, 2011 @04:30PM (#38151918) Homepage Journal

    People in Kansas know that the Debian Package Dependencies were Intelligently Designed!!!

    • by CAIMLAS (41445)

      There's no evidence that both intelligent design and evolution is involved, in the biological case. After all, Debian, which is designed and maintained by intelligent beings, demonstrates a very strong similarity to evolutionary process. It's not a proof in and of itself, but it's suggestive, IMO.

    • God said "sudo apt-get update" and God saw the repositories were good and God said "sudo apt-get upgrade" ... and on the seventh day God was able to connect by wifi to his local hotspot after finally getting his ndiswrapper drivers working and he rested.
  • by hyades1 (1149581) <hyades1@hotmail.com> on Wednesday November 23, 2011 @04:44PM (#38152072)

    I sometimes suspect that my former marriage could have been successfully modeled as a predator/prey relationship.

  • by ortholattice (175065) on Wednesday November 23, 2011 @06:16PM (#38153178)
    I don't have access to the full version. Maybe I overlooked something, but I couldn't find anything in the abstract or article that gave more than a fuzzy picture about the "relationship" they are discussing. The set of equations normally used are the Lotka-Volterra equations [wikipedia.org]. An example of the behavior of this equation is shown here [wikipedia.org] on that page. Is this what they found matches their data?

Do molecular biologists wear designer genes?

Working...