Re: 2.3.51 tulip broken

From: Dave_Pfaltzgraff@patapsco.com
Date: Mon Mar 20 2000 - 16:54:37 EST

  • Next message: Michael Harnois: "[PATCH] tulip hosed in pre3-3"

    Dave Pfaltzgraff@PATAPSCO
    03/20/2000 04:54 PM

    Until now, I was a curious bystander to this thread, but now I feel compelled to
    put in my two cents. Acknowledging that some may feel that two cents is not much
    value these days, I still feel that there is significance in what I have to say.
    Unfortunately, I know of no other podium where those who may benefit the most
    may be listening. Please hear me out.

    This thread has often drawn the analogy to Eric's "The Cathedral and the
    Bazaar". If you have ever experienced the true bazaar environment, you come away
    feeling that everything is chaotic and it's a wonder anything ever gets done.
    This is contrasted with the cathedral where it seems that order reigns to the
    detriment of progress. Admittedly, this focuses on the two extremes. However,
    there are many variations in between that work well and it would seem that most
    companies are somewhere in between on the spectrum. Linux is maturing and a part
    of the growth process is to re-evaluate where we've been, where we are now and
    where we are headed. It also gives us the chance to re-evaluate where we want to
    be in this operational spectrum.

    From my viewpoint, I would like to see Linux continue on the path of being a
    full featured and reliable operating system. (This contrasts with those who
    think it should be a 'David' to Microsoft's 'giant'.) In order to be both full
    featured and reliable, the Linux community must continue to draw on an ever
    increasing base of developers. This places more demands on the current
    developers and emphasizes the need for better communications. (I'll come back to
    this shortly.)

    In contrast to my viewpoint, there is another group that would like to see Linux
    support all of the 'latest and greatest'. Whether that's hardware or software
    technology, it doesn't matter. This group emphasizes the speed of development.
    (Just look at the pressure to release 2.4 by some arbitrary date! Was that
    pressure put there by some imaginary 'competitor'?) I hate to say it, but this
    environment emphasizes what has been called the 'cowboy' development - act now,
    talk about it (document) later. Yes, there's room and actually a need for this
    type of environment, but to anyone not directly involved, the feeling is one of
    complete chaos - the bazaar. Also, 'document later' has the effect of
    encouraging important details to get lost in the shuffle.

    Having been a student of computer technology (for several decades), I have
    learned the need for clearly defined interfaces. This is true whether you're
    speaking of hardware, software, user interfaces or almost any place where you
    can draw a line between levels of interaction and/or responsibility. Not knowing
    the inner workings of the Linux kernel at this point, I have to raise the
    question, as some others have in this thread, as to why the network is
    integrated into the kernel to the degree implied by those who are crying that
    the Tulip driver is broken. I can compile a kernel without a network and have
    done it many times for embedded systems. This confirms my feeling that the
    network is an interface supported by the kernel and not necessarily an integral
    part it. (Sort of reminds me of Microsoft and the integrated browser arguments!)

    Back to the topic of the need for better communications... It would appear to me
    that the group of developers working on the kernel should unambiguously define
    an interface and allow development efforts to proceed in parallel. If there is
    to be change in that interface, then both parties must agree to it. In any
    development, it is extremely important for all participants to 'buy into' the
    design. From what I have read here, it would appear to me that this has not been
    the case. Under these circumstances, it's only natural that finger pointing
    should begin and I'll admit to having been amused by some of what has been said
    on both sides.

    What I'd like to say now is merely: Come on guys (and gals, as the case may be)!
    Let's grow up and accept responsibility for what has happened and for what's
    going on now. This kind of rift is exactly what will destroy the momentum of
    Linux. Momentum is not measured by release numbers or rates. It's measured by
    the size of the community involved with Linux - both developers and users. Let
    us all acknowledge that our present development course is on the verge of
    floundering and ask if there is some thing that we can do to get over this and
    get moving together again. (It's my sincere hope that the rift is not
    irreparable.) Along with this, let us acknowledge that for some the bazaar
    environment is exactly what they need while for others the cathedral environment
    works better. Let's discuss not who needs to do what to which module, but how
    are we going to move together into a more open (accepting) manner that will
    benefit us all. This is a community of developers willing to donate their time
    and efforts. In this environment, coercion will not work. Only cooperation will
    produce results.

    With the best of intentions and hopes for the future,
    Dave

    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.rutgers.edu
    Please read the FAQ at http://www.tux.org/lkml/



    This archive was generated by hypermail 2b29 : Mon Mar 20 2000 - 19:30:56 EST