real network cards - Minix

This is a discussion on real network cards - Minix ; hello i have three things in mind i'd like to do: - run gnustep on minix. - make a debian gnu/minix port of it. - create a livecd of a gnustep based system with minix. however i failed to get ...

+ Reply to Thread
Results 1 to 9 of 9

Thread: real network cards

  1. real network cards

    hello i have three things in mind i'd like to do:

    - run gnustep on minix.

    - make a debian gnu/minix port of it.

    - create a livecd of a gnustep based system with minix.

    however i failed to get any of nic i have to run with minix3... is
    there
    a list of supported nics (with detailed pci id of them)?


  2. Re: real network cards

    It's rather simple.

    What you are hoping to achive cannot be achived in this decade. I'll be
    frank -- Minix has over 10 years of development to go before it gets
    even close to being able of doing what you want. GNUstep has a hard
    requirement on shared libraries. Look at Minix 3, it still uses a.out
    format statically linked binaries -- something most UNIX vendors gave up
    15 years ago. Not to mention that our X server is anything but fast, and
    that the supported NIC list is basically 5 varieties of cards.

    That is:
    NE2000-PCI and compatable (Read the fscking Linux driver help on this,
    Realtek RTL8029's and NE34-PCIs are clones)
    Whatever VMWare emulates (I can;'t remember off thetop of my head)
    Realtek RTL8139
    Intel PRO 100 PCI (I have one of these cards that the driver identifies
    and acknoldges, but also flat out refuses to work with, possibly a odd
    revision, note that the driver name is 'fxp' in both Minix 3 and
    FreeBSD)
    3COM 3c509 "EtherLink III PnP" (ISAPnP)

    Drop your fruitless quest for now. I think you might want to try again
    in the year 2016, Minix 3 *might* be ready by then. (Sadly, this isn't a
    troll, but more of a realistic projection)

    silverd@gmail.com wrote:
    > hello i have three things in mind i'd like to do:
    >
    > - run gnustep on minix.
    >
    > - make a debian gnu/minix port of it.
    >
    > - create a livecd of a gnustep based system with minix.
    >
    > however i failed to get any of nic i have to run with minix3... is
    > there
    > a list of supported nics (with detailed pci id of them)?
    >




    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.1 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFFHfyNsjeOFtd+nycRAhlfAJ9GivYRkm6sjbWm3LbE/A6ZgfFxBgCeJgpo
    9oz393HtNWLIhT7k92vP3sg=
    =ePQn
    -----END PGP SIGNATURE-----


  3. Re: real network cards

    > It's rather simple.
    >
    > What you are hoping to achive cannot be achived in this decade. I'll be
    > frank -- Minix has over 10 years of development to go before it gets
    > even close to being able of doing what you want. GNUstep has a hard
    > requirement on shared libraries.


    Thank you for being so informative. That sure saves me a lot of time.
    So I'll stick with Debian GNU/Linux and FreeBSD for now. But the
    longer term goal sure is a microkernel I think. Maybe GNU (Hurd)
    soon will support sound cards.

    > Look at Minix 3, it still uses a.out
    > format statically linked binaries -- something most UNIX vendors gave up
    > 15 years ago. Not to mention that our X server is anything but fast, and
    > that the supported NIC list is basically 5 varieties of cards.
    > That is:
    > NE2000-PCI and compatable (Read the fscking Linux driver help on this,
    > Realtek RTL8029's and NE34-PCIs are clones)
    > Whatever VMWare emulates (I can;'t remember off thetop of my head)
    > Realtek RTL8139
    > Intel PRO 100 PCI (I have one of these cards that the driver identifies
    > and acknoldges, but also flat out refuses to work with, possibly a odd
    > revision, note that the driver name is 'fxp' in both Minix 3 and
    > FreeBSD)
    > 3COM 3c509 "EtherLink III PnP" (ISAPnP)


    I had the same Intel PCI id driver problem.

    > Drop your fruitless quest for now. I think you might want to try again
    > in the year 2016, Minix 3 *might* be ready by then. (Sadly, this isn't a
    > troll, but more of a realistic projection)


    Your answer was very professional, I've never had the thought of
    you being a troll. Thank you for the answer, the idea got dropped.
    Maybe it will change in a decade of years, maybe it won't. Maybe
    we'll see (or not). Have a nice time


  4. Re: real network cards


    silverd@gmail.com writes:

    > > It's rather simple.
    > >
    > > What you are hoping to achive cannot be achived in this decade. I'll be


    Segin, who so often bashes other peoples grasp of English should note,
    that achieve is written with an 'e' after the 'i'.

    > > frank -- Minix has over 10 years of development to go before it gets
    > > even close to being able of doing what you want. GNUstep has a hard
    > > requirement on shared libraries.

    >
    > Thank you for being so informative. That sure saves me a lot of time.


    I'd like to add that this is only Segin's rather biased opinion.
    Don't believe him. For reasons I cannot comprehend completely, Segin
    seems to have some aversions to the present state of "the Minix
    project" or it's "management". So he is prone to exaggerate his
    position on this.

    > So I'll stick with Debian GNU/Linux and FreeBSD for now. But the
    > longer term goal sure is a microkernel I think. Maybe GNU (Hurd)
    > soon will support sound cards.


    AFAIS Debian Gnu/Minix is still only in the planning stages (sorry to
    the Debian Gnu/Minix activists if I'm mistaken). Linux and the BSDs
    have today what you need. So yes, if you want to get something done
    now which needs features not yet in Minix3, you're probably better of
    doing it at one of these systems. (But if I understood ASTs answer to
    Debian Gnu/Minix right, the Minix people are always rather interested
    to know what is missing in Minix that prevents you from porting your
    stuff to Minix).

    As far as the Minix3 state goes, as someone not affiliated with the
    project or the team at VU, I don't share Segin's position. What I now
    say is the impression gathered from following this group (c.o.m) and
    doing a bit of web search now and then:

    - VM support is nearly complete.

    - Shared library support is in the works and the next big thing to come.

    - There have been a discussion on sound card and NIC drive frameworks
    quite recently. Since Minix3 is actively developed now, I'd expect
    that a wider support of cards will happen not really soon, but within
    1 or 2 years.

    I will probably not use Minix for my _work_ in the next years, but
    neither do I share Segin's bleak outlook.

    Minix3 is coming along nicely and will become a lean hobbyist system
    or an OS for niche applications (where you either select the hardware
    after the software or write your own device drivers).

    But I wouldn't expect a really all encompassing hardware driver
    support even in the next 5-10 years, considering that even Linux has
    (again) increasing problems with that, which, I hasten to add, are all
    caused by hardware vendors who mistakenly think that they have to
    protect or cannot document the software/hardware interface (like ATI,
    NVidia, Fritz! etc). In their mind set, windows and the PC hardware
    form a single unit. They are only taking up Linux support very
    reluctantly (with binary drivers) and will never support Minix. Since
    the interface is undocumented -- how can anybody write drivers for
    that hardware?

    > > Look at Minix 3, it still uses a.out format statically linked
    > > binaries -- something most UNIX vendors gave up 15 years ago.


    That is a criterion for usefulness or whatever? That most of the other
    Unix vendors went out of business has of course nothing to do with
    their choice of binary format but I'm mightily tempted to just retort:
    "Yeah, and that is exactly why we have so much of the old UNIX vendors
    in the market today!".

    > > NE2000-PCI and compatable (Read the fscking Linux driver help on this,


    <...>

    > > Drop your fruitless quest for now. I think you might want to try again
    > > in the year 2016, Minix 3 *might* be ready by then. (Sadly, this isn't a
    > > troll, but more of a realistic projection)


    > Your answer was very professional, I've never had the thought of


    No it wasn't. "fscking" in example, is not a technical term. It was a
    typical Segin answer: Plenty opiniated, not totally wrong or off
    topic, but somehow quite off center.

    > you being a troll.


    I also cannot imagine how that idea might ever have cropped up.

    > Thank you for the answer, the idea got dropped.


    Dropping the idea for now might have been the right decision for you
    now. But hopefully not for the wrong reasons. :-)

    Regards -- Markus


  5. Re: real network cards

    OK, you flabbergasted troll, like you yourself said, there's MUCH work
    to be done before half of this is even possible. I'll keep it
    short-and-sweet, but talk is just that -- talk. The shared library
    support and complete VM you speak of are worth nothing more than you say
    my statements are.

    And by the way, did I mention that's the context switching penalty
    common of microkernel systems?

    M E Leypold wrote:
    > silverd@gmail.com writes:
    >
    >>> It's rather simple.
    >>>
    >>> What you are hoping to achive cannot be achived in this decade. I'll be

    >
    > Segin, who so often bashes other peoples grasp of English should note,
    > that achieve is written with an 'e' after the 'i'.
    >
    >>> frank -- Minix has over 10 years of development to go before it gets
    >>> even close to being able of doing what you want. GNUstep has a hard
    >>> requirement on shared libraries.

    >> Thank you for being so informative. That sure saves me a lot of time.

    >
    > I'd like to add that this is only Segin's rather biased opinion.
    > Don't believe him. For reasons I cannot comprehend completely, Segin
    > seems to have some aversions to the present state of "the Minix
    > project" or it's "management". So he is prone to exaggerate his
    > position on this.
    >
    >> So I'll stick with Debian GNU/Linux and FreeBSD for now. But the
    >> longer term goal sure is a microkernel I think. Maybe GNU (Hurd)
    >> soon will support sound cards.

    >
    > AFAIS Debian Gnu/Minix is still only in the planning stages (sorry to
    > the Debian Gnu/Minix activists if I'm mistaken). Linux and the BSDs
    > have today what you need. So yes, if you want to get something done
    > now which needs features not yet in Minix3, you're probably better of
    > doing it at one of these systems. (But if I understood ASTs answer to
    > Debian Gnu/Minix right, the Minix people are always rather interested
    > to know what is missing in Minix that prevents you from porting your
    > stuff to Minix).
    >
    > As far as the Minix3 state goes, as someone not affiliated with the
    > project or the team at VU, I don't share Segin's position. What I now
    > say is the impression gathered from following this group (c.o.m) and
    > doing a bit of web search now and then:
    >
    > - VM support is nearly complete.
    >
    > - Shared library support is in the works and the next big thing to come.
    >
    > - There have been a discussion on sound card and NIC drive frameworks
    > quite recently. Since Minix3 is actively developed now, I'd expect
    > that a wider support of cards will happen not really soon, but within
    > 1 or 2 years.
    >
    > I will probably not use Minix for my _work_ in the next years, but
    > neither do I share Segin's bleak outlook.
    >
    > Minix3 is coming along nicely and will become a lean hobbyist system
    > or an OS for niche applications (where you either select the hardware
    > after the software or write your own device drivers).
    >
    > But I wouldn't expect a really all encompassing hardware driver
    > support even in the next 5-10 years, considering that even Linux has
    > (again) increasing problems with that, which, I hasten to add, are all
    > caused by hardware vendors who mistakenly think that they have to
    > protect or cannot document the software/hardware interface (like ATI,
    > NVidia, Fritz! etc). In their mind set, windows and the PC hardware
    > form a single unit. They are only taking up Linux support very
    > reluctantly (with binary drivers) and will never support Minix. Since
    > the interface is undocumented -- how can anybody write drivers for
    > that hardware?
    >
    >>> Look at Minix 3, it still uses a.out format statically linked
    >>> binaries -- something most UNIX vendors gave up 15 years ago.

    >
    > That is a criterion for usefulness or whatever? That most of the other
    > Unix vendors went out of business has of course nothing to do with
    > their choice of binary format but I'm mightily tempted to just retort:
    > "Yeah, and that is exactly why we have so much of the old UNIX vendors
    > in the market today!".
    >
    >>> NE2000-PCI and compatable (Read the fscking Linux driver help on this,

    >
    > <...>
    >
    >>> Drop your fruitless quest for now. I think you might want to try again
    >>> in the year 2016, Minix 3 *might* be ready by then. (Sadly, this isn't a
    >>> troll, but more of a realistic projection)

    >
    >> Your answer was very professional, I've never had the thought of

    >
    > No it wasn't. "fscking" in example, is not a technical term. It was a
    > typical Segin answer: Plenty opiniated, not totally wrong or off
    > topic, but somehow quite off center.
    >
    >> you being a troll.

    >
    > I also cannot imagine how that idea might ever have cropped up.
    >
    >> Thank you for the answer, the idea got dropped.

    >
    > Dropping the idea for now might have been the right decision for you
    > now. But hopefully not for the wrong reasons. :-)
    >
    > Regards -- Markus
    >




    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.1 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFFIXa3sjeOFtd+nycRAgf0AJ9aTqxx3e1136ius7NCz7/wHGIaMwCfUyeY
    lbQrxjVgKf0fBpeWxfFIQW4=
    =eHik
    -----END PGP SIGNATURE-----


  6. Re: real network cards


    Segin writes:

    > OK, you flabbergasted troll, like you yourself said, there's MUCH


    According to my Oxford dictionary, "flabbergasted" (US) means "to
    overwhelm with amazement". So do you mean I'm overwhelmed with
    amazement? Or did you mean "flabbergasting" (I.e. I'm overwhelming you
    with amazement)? Else it doesn't seem to make much sense, since I'm
    not overwhelmed by your little speech about the multiple and manifold
    failings of Minix 3. (Well, perhaps I'm amazed ...).

    > work to be done before half of this is even possible.


    > I'll keep it short-and-sweet, but talk is just that -- talk.


    Right. What, BTW, did you do, to advance Minix? Me, I admit that I'm
    only a former Minix 2 user with some technical interests hanging
    around at c.o.m and looking what's happening with Minix 3. Not much
    ambition there.

    You, on the other side, you disdain idle talking, you seem to know
    very well, what's wrong with Minix and where the project management
    failed (if I understood one of your last missives correctly). One
    word: You're a man of action. Why don't you fork and do everything
    better than the VU team and the volunteers that have contributed so
    far?

    Of course you might end up doing it alone ... - and considering that
    you learn something about the social aspects of software developments,
    of "maintaining the development process", which is specially important
    in open source projects. It is probably not enough to have an attitude
    like Theo de Raadt, but you need a convincing technical vision also --
    to counterbalance the attitude.

    > The shared library support and complete VM you speak of are worth
    > nothing more than you say my statements are.


    Well, maybe. We will just see. I'd expected since somebody has written
    his thesis on the VM that it would be already fairly complete and
    working -- as opposed to "worthless". As far as the shared library
    support goes, you must have better intelligence on project internals
    than I do, since you know so sure that it will come to nothing for
    years.

    > And by the way, did I mention that's the context switching penalty
    > common of microkernel systems?


    I unfortunately fail to make any sense from the last sentence. What
    does the "that" refer to?

    You never fail to amaze me :-). And persistently top posting also: If
    that is not a sure sign that you must have diamond teeth, I don't
    know.

    Regards -- Markus


    > M E Leypold wrote:
    > > silverd@gmail.com writes:




  7. Re: real network cards

    M E Leypold wrote:
    > Right. What, BTW, did you do, to advance Minix? Me, I admit that I'm
    > only a former Minix 2 user with some technical interests hanging
    > around at c.o.m and looking what's happening with Minix 3. Not much
    > ambition there.
    >


    I've ported many programs, wget, nano, webcpp, all that was me. That
    there isn't the entire list of what I've done, but it's not the end of
    what I plan on doing.

    And I plan on adding more contributions in due time, although right now
    I am kinda stuck (I fought with parted and parted won...)

    And if you ever need help with porting programs (autoconf or not), I
    will be glad to cut the bull**** and give a hand.


    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.1 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFFIzyusjeOFtd+nycRAkZgAJ9ToS+NVjDgqUhcJdHeJ0 IW/44xjgCdE1RL
    I7HJLvxuLkOWyhc7T8VZIsA=
    =slVE
    -----END PGP SIGNATURE-----


  8. Re: real network cards

    M E Leypold wrote:
    > silverd@gmail.com writes:
    > But I wouldn't expect a really all encompassing hardware driver
    > support even in the next 5-10 years, considering that even Linux has
    > (again) increasing problems with that, which, I hasten to add, are all
    > caused by hardware vendors who mistakenly think that they have to
    > protect or cannot document the software/hardware interface (like ATI,
    > NVidia, Fritz! etc). In their mind set, windows and the PC hardware
    > form a single unit. They are only taking up Linux support very
    > reluctantly (with binary drivers) and will never support Minix. Since
    > the interface is undocumented -- how can anybody write drivers for
    > that hardware?

    This is my crack pot idea (I'm really fairly new to Operating systems,
    so bear with me): since Minix 3 can work without access to some
    drivers, couldn't it be feasible to write a server/daemon/whatever that
    writes a new driver?

    Say it keeps track of the drivers and when a driver performance is god
    awful (it crashes some number of times within some time period), it
    simply sees this, takes it out of operation, and (idealistically) uses
    evolution to write a new driver, compiles it, then loads it up.

    Now if this hosh-posh of crack pot idealism works, you wouldn't need to
    write drivers manually. Of course, this server/daemon/whatever would be
    so complicated it would probably need to take 5-10 years to get it
    running perfectly.


  9. Re: real network cards

    Pablo Rodriguez wrote:
    > M E Leypold wrote:

    [snip]
    > This is my crack pot idea (I'm really fairly new to Operating systems,
    > so bear with me): since Minix 3 can work without access to some
    > drivers, couldn't it be feasible to write a server/daemon/whatever that
    > writes a new driver?
    >
    > Say it keeps track of the drivers and when a driver performance is god
    > awful (it crashes some number of times within some time period), it
    > simply sees this, takes it out of operation, and (idealistically) uses
    > evolution to write a new driver, compiles it, then loads it up.
    >
    > Now if this hosh-posh of crack pot idealism works, you wouldn't need to
    > write drivers manually. Of course, this server/daemon/whatever would be
    > so complicated it would probably need to take 5-10 years to get it
    > running perfectly.


    Excuse me, but i need to object to the term "evolution" being used
    here;

    Pablo, what you're really referring to would be better classified and
    named as
    ``successive, automatic refinement of programm.''

    Evilution, such as it is, does not and cannot refer to dead things.

    It's the same kind of incorrect usage as Einstein's relativity being
    applied to
    non-physics situations which he vehemently dealt with after he authored
    his paper
    on Relativity until the end of his life.

    thank you.


+ Reply to Thread