Minix = Just another Free UNIX? - Minix

This is a discussion on Minix = Just another Free UNIX? - Minix ; I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms. They are so hard to grasp! For example, signals. Who needs 'em if you have a nice clean event model? And who forks anymore? That ...

+ Reply to Thread
Results 1 to 13 of 13

Thread: Minix = Just another Free UNIX?

  1. Minix = Just another Free UNIX?

    I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    They are so hard to grasp! For example, signals. Who needs 'em if you have a
    nice clean event model? And who forks anymore? That is soooo yesterday.

    My point is this: Minix shouldn't be "competing" with free UNIX/Linux. It
    should be bold and jetison tradition and conformity when beneficial: it
    should one-up Windows and Mac!

    John



  2. Re: Minix = Just another Free UNIX?


    "JohnQ" writes:

    > I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    > They are so hard to grasp! For example, signals. Who needs 'em if you have a
    > nice clean event model? And who forks anymore? That is soooo yesterday.


    Actually fork/exec and inheritance to the exec'd process is quite a
    smart model to set up new processes. You'll see that when comparing with
    CreateProcess and its 1000 flags.

    Fork'ing is also a good alternative to a full thread model, see the
    way, Apache and other servers handle this. The advantage is, that the
    isolation between quasi-threads is greater and that the accumulated
    entropy from bugs (e.g. slow heap corruption or leaks) is ditched when
    the worker processes die after processing N requests or units of work.

    This is also evidenced by the better stability of fork/exec model
    servers if compared to purely threaded server (where every bit of
    memory corruption will potentially stay to the end of the server
    processes uptime).

    > My point is this: Minix shouldn't be "competing" with free UNIX/Linux.


    It doesn't. BTW the fork/exec model wasn't new in Unix. It has been
    used in GENIE before (AFAIR, I'm toooo lazy to look it up now).

    > It should be bold and jetison tradition and conformity when
    > beneficial: it should one-up Windows and Mac!


    Why? I hope you're not trolling only and can suggest a useful
    alternative API. From what you write POSIX is out and Win32 too. Which
    API do you suggest? And about which APIs are you talking? Files,
    process creation and control, any kind of I/O to user etc.

    Might Minix support the ANSI C library?

    And back to the events you quote in your third sentence: What would
    you do if a async event arrives in a process? Which API for async
    event handling would you suggest?

    Regards -- Markus



  3. Re: Minix = Just another Free UNIX?

    JohnQ wrote:

    > I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    > They are so hard to grasp! For example, signals. Who needs 'em if you have
    > a nice clean event model? And who forks anymore? That is soooo yesterday.
    >
    > My point is this: Minix shouldn't be "competing" with free UNIX/Linux. It
    > should be bold and jetison tradition and conformity when beneficial: it
    > should one-up Windows and Mac!
    >
    > John


    I think the word is "Put up or shut up." If you know the Win32 API and
    consider it a good-enough API, and dislike the POSIX-style, then nobody's
    stopping you from writing an OS that 'one-ups' Windows, and does it better
    than Microsoft can. After all, Atheos was designed around a combination of
    the Amiga OS API and the BeOS API, and its fork, Syllable, is one of the
    most promising new OSes.

    Minix has been around and was designed to be a Unix-class OS, without many
    of the problems Unix have accumulated. It's not going to change just on
    your say-so.

    So, you've got a challenge now - 'one-up' Microsoft Windows and let us know.

  4. Re: Minix = Just another Free UNIX?


    "Tux Wonder-Dog" wrote in message
    news:45f12b35@clear.net.nz...
    > JohnQ wrote:
    >
    >> I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    >> They are so hard to grasp! For example, signals. Who needs 'em if you
    >> have
    >> a nice clean event model? And who forks anymore? That is soooo yesterday.
    >>
    >> My point is this: Minix shouldn't be "competing" with free UNIX/Linux. It
    >> should be bold and jetison tradition and conformity when beneficial: it
    >> should one-up Windows and Mac!
    >>
    >> John

    >
    > I think the word is "Put up or shut up." If you know the Win32 API and
    > consider it a good-enough API, and dislike the POSIX-style, then nobody's
    > stopping you from writing an OS that 'one-ups' Windows, and does it better
    > than Microsoft can. After all, Atheos was designed around a combination
    > of
    > the Amiga OS API and the BeOS API, and its fork, Syllable, is one of the
    > most promising new OSes.
    >
    > Minix has been around and was designed to be a Unix-class OS, without many
    > of the problems Unix have accumulated. It's not going to change just on
    > your say-so.
    >
    > So, you've got a challenge now - 'one-up' Microsoft Windows and let us
    > know.


    I would (and personally, I think it is pretty easy to do so), but
    unfortunately, I don't have that much time left to live nor do I have the
    resources to do so. My question though was to prompt consideration or get
    feedback on what the "plan" is for Minix. You seem to be confirming that
    Minix is going to pretty much be yet another UNIX-like OS (but I don't know
    who the "powers to be" are that make or will make or have made that decision
    are). At this early juncture in its evolution, I was wondering if (or
    suggesting that) Minix could have a greater ambition/potential. Basicly
    Windows and UNIX both suck, but have their their merits. Synthesizing (or
    simply learning from) those could be pivotal. Maybe OS X has already done
    that? Or Open Solaris?

    Just a thought.

    (BTW, your attitude really sucks).

    John



  5. Re: Minix = Just another Free UNIX?

    On Mar 4, 3:11 am, "JohnQ"
    wrote:
    > I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    > They are so hard to grasp! For example, signals. Who needs 'em if you have a
    > nice clean event model? And who forks anymore? That is soooo yesterday.
    >
    > My point is this: Minix shouldn't be "competing" with free UNIX/Linux. It
    > should be bold and jetison tradition and conformity when beneficial: it
    > should one-up Windows and Mac!
    >
    > John


    The goal of Minix is to provide *robust* operating system. There are
    things to improve (this goal is ambitious) but In this respect they
    are already ahead of Linux/Windows (I am not sure about Mac kernel).

    (my view)


  6. Re: Minix = Just another Free UNIX?

    Hello,

    Minix was created in order to provide a FREE (without much cost) UNIX-like
    operating system for teaching. You cannot take out the Unix of that.

    Minix is older than Linux, and is still meant to teaching. If it nowadays
    works as expected and provides a unix-like POSIX-API it only can help. If
    you want to invent another API, you can do it. But porting/writing
    applications this way may not be as easy.

    The complete PC architecture is based on "compatibility", going as far as
    introducing fixed bugs (UART and XT's BIOS) or having different modes of
    addressing the memory (real, protected, virtual86, long mode) only for being
    able to run DOS 3.x (which still works on AMD64, although Win 3.x is said
    only to work in Standard Mode).

    If you want something "really" good, then invent some better architecture,
    make it usable and port Windows onto it. Only then it will find acceptance
    on users. Other people (devs) don't care if the API isn't good enough - they
    simply use it or make it better.

    Just my 0.02$.

    Regards



  7. Re: Minix = Just another Free UNIX?


    "JohnQ" writes:

    > "Tux Wonder-Dog" wrote in message
    > news:45f12b35@clear.net.nz...
    >> JohnQ wrote:
    >>
    >>> I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    >>> They are so hard to grasp! For example, signals. Who needs 'em if you
    >>> have
    >>> a nice clean event model? And who forks anymore? That is soooo yesterday.
    >>>
    >>> My point is this: Minix shouldn't be "competing" with free UNIX/Linux. It
    >>> should be bold and jetison tradition and conformity when beneficial: it
    >>> should one-up Windows and Mac!
    >>>
    >>> John

    >>
    >> I think the word is "Put up or shut up." If you know the Win32 API and
    >> consider it a good-enough API, and dislike the POSIX-style, then nobody's
    >> stopping you from writing an OS that 'one-ups' Windows, and does it better
    >> than Microsoft can. After all, Atheos was designed around a combination
    >> of
    >> the Amiga OS API and the BeOS API, and its fork, Syllable, is one of the
    >> most promising new OSes.
    >>
    >> Minix has been around and was designed to be a Unix-class OS, without many
    >> of the problems Unix have accumulated. It's not going to change just on
    >> your say-so.
    >>
    >> So, you've got a challenge now - 'one-up' Microsoft Windows and let us
    >> know.

    >
    > I would (and personally, I think it is pretty easy to do so), but
    > unfortunately, I don't have that much time left to live nor do I have the
    > resources to do so.


    "Not having the time or the ressources" is actually the proper
    definition of "not so easy". Furthermore consider wether the
    project should become an exercise for personal gratification or
    something other people use: Windows got "one-upped" often enough (by
    Unix more than once in my count, also by OS/2), Unix got "one-upped"
    by Plan9 and Amoeba. Is anybody using those system? If not, what are
    the reasons? Actually this illustrates very well, that technical
    superiority without (enough, mental/conceptual) compatibility will
    leave a new system out in the cold.

    > My question though was to prompt consideration or get
    > feedback on what the "plan" is for Minix.


    Question? It was a call to change things and to deviate from proven
    good practice. Without you providing any good alternative.

    Let me repeat my questions: What is your alternative API design?

    > You seem to be confirming that Minix is going to pretty much be yet
    > another UNIX-like OS (but I don't know who the "powers to be" are
    > that make or will make or have made that decision are).


    Is already another UNIX-like OS. Has been for some umpteenth years.

    > At this early juncture in its evolution, I was wondering if (or


    Which early junctions? Are you sure you're not reading Google archives
    from somewhere around the end of the 80s?

    > suggesting that) Minix could have a greater
    > ambition/potential. Basicly Windows and UNIX both suck, but have
    > their their merits. Synthesizing (or simply learning from) those
    > could be pivotal.


    What exactly do you suggest? Could you provide _some_ more details,
    please?

    > Maybe OS X has already done that? Or Open Solaris?


    So you don't know? Why don't you find out?

    > Just a thought.


    Well, think some more.

    > (BTW, your attitude really sucks).


    Mine too.

    Regards -- Markus

  8. Re: Minix = Just another Free UNIX?

    JohnQ wrote:

    >
    > "Tux Wonder-Dog" wrote in message
    > news:45f12b35@clear.net.nz...
    >> JohnQ wrote:
    >>
    >>> I hope not. I'm from the Windows world of APIs so I abhor the UNIX-isms.
    >>> They are so hard to grasp! For example, signals. Who needs 'em if you
    >>> have
    >>> a nice clean event model? And who forks anymore? That is soooo
    >>> yesterday.
    >>>
    >>> My point is this: Minix shouldn't be "competing" with free UNIX/Linux.
    >>> It should be bold and jetison tradition and conformity when beneficial:
    >>> it should one-up Windows and Mac!
    >>>
    >>> John

    >>
    >> I think the word is "Put up or shut up." If you know the Win32 API and
    >> consider it a good-enough API, and dislike the POSIX-style, then nobody's
    >> stopping you from writing an OS that 'one-ups' Windows, and does it
    >> better
    >> than Microsoft can. After all, Atheos was designed around a combination
    >> of
    >> the Amiga OS API and the BeOS API, and its fork, Syllable, is one of the
    >> most promising new OSes.
    >>
    >> Minix has been around and was designed to be a Unix-class OS, without
    >> many
    >> of the problems Unix have accumulated. It's not going to change just on
    >> your say-so.
    >>
    >> So, you've got a challenge now - 'one-up' Microsoft Windows and let us
    >> know.

    >
    > I would (and personally, I think it is pretty easy to do so), but
    > unfortunately, I don't have that much time left to live nor do I have the


    Might I suggest you start the project, release it under a
    well-known-and-understood Free/Libre and Open Source license such as the
    GPL or the BSD license, tell people about it, get it set up working with an
    online community such as the one Linus Torvalds has set up, and then your
    personal TTL won't be such a problem.

    It's not as if there aren't F/LOSS people who know the Win32 API and prefer
    it to the POSIX one - there are, and some of them probably think they could
    do better without its historical baggage. Such as Win16 compatibility.

    > resources to do so. My question though was to prompt consideration or get
    > feedback on what the "plan" is for Minix. You seem to be confirming that
    > Minix is going to pretty much be yet another UNIX-like OS (but I don't
    > know who the "powers to be" are that make or will make or have made that
    > decision are). At this early juncture in its evolution, I was wondering if
    > (or suggesting that) Minix could have a greater ambition/potential.


    Minix was initially a teaching system. It came as part of a book for
    students, and was sold as a distribution independently of the book.

    It wasn't until well past the Minix 2.x stage that I, for one, became aware
    that there was a Minix 3.x in the wings, and by that time, Minix was
    well-known as a "miniature Unix" - that's what the name means.

    Since it's now released under the BSD license, I myself could no doubt do
    what you suggest and roll a completely different "Win32-ized" version of
    it, without having to pester Andy Tanenbaum or the Minix community about
    it. But what I couldn't do is call it Minix, for one thing - for example,
    there's a VMS hobbyist who's doing something of the sort with a fork of the
    Linux 2.4 kernel, and he calls his fork FreeVMS.

    > Basicly Windows and UNIX both suck, but have their their merits.
    > Synthesizing (or simply learning from) those could be pivotal. Maybe OS X
    > has already done that? Or Open Solaris?


    At the very least, write a document on the benefits and mistakes of the
    Win32 API and the POSIX API and what could and should be done better, put
    it up on the interweb and post references to it on various OS newsgroups.

    That's more constructive than merely complaining about their weaknesses and
    wondering if some community whose interest is firmly into getting their
    particular OS working correctly, should do as you suggest.
    >
    > Just a thought.
    >
    > (BTW, your attitude really sucks).


    I know. Sad, isn't it!
    >
    > John


    Wesley Parish

  9. Re: Minix = Just another Free UNIX?


    "Tux Wonder-Dog" wrote in message
    news:45f510d8@clear.net.nz...

    > Since it's now released under the BSD license, I myself could no doubt do
    > what you suggest and roll a completely different "Win32-ized" version of
    > it, without having to pester Andy Tanenbaum or the Minix community about
    > it. But what I couldn't do is call it Minix, for one thing - for example,
    > there's a VMS hobbyist who's doing something of the sort with a fork of
    > the
    > Linux 2.4 kernel, and he calls his fork FreeVMS.


    I'll have to look at the source to see to what degree UNIX-isms are in the
    core code to see if forking is viable. That's actually what I was getting at
    from the get go though: having a neutral foundation at the kernal level upon
    which to build whatever you want without having to ascribe to a certain
    model. If it's too UNIX-y, I'm not going to like it. The question would
    become, can I change it or not. I haven't worked on anything that low level,
    and I don't think I want to anyway and subject myself to working in parallel
    with the Minix team (a waste of effort).

    ("Miniature UNIX"... duh! I could've had a V8!).

    John



  10. Re: Minix = Just another Free UNIX?


    "Tux Wonder-Dog" wrote in message
    news:45f510d8@clear.net.nz...

    > It wasn't until well past the Minix 2.x stage that I, for one, became
    > aware
    > that there was a Minix 3.x in the wings, and by that time, Minix was
    > well-known as a "miniature Unix" - that's what the name means.


    Ahhh yes! That seems to have escaped me somehow.

    John



  11. Re: Minix = Just another Free UNIX?

    On Mon, 12 Mar 2007 16:36:48 -0500, JohnQ wrote:

    > having a neutral foundation at the kernal level upon
    > which to build whatever you want without having to ascribe to a certain
    > model. If it's too UNIX-y, I'm not going to like it.


    From memory of reading the code pre 1.5-era, Minix is pretty Unix-y all
    the way down. (Unix V7!) All you've got is processes (CSP-style). If
    you want anything that speaks threads or even lower-level, then you'll
    need to start from Mach or L4 (or something else).

    What have you got against posix? It's simple and it works.

    You could look into the iTron flavour for eCOS, if you're after something
    different. Don't know how up-to-date that is, though. I think that the
    posix API gets most of the running.

    Cheers,

    --
    Andrew


  12. Re: Minix = Just another Free UNIX?


    "Andrew Reilly" wrote in message
    newsan.2007.03.13.14.03.57.246536@areilly.bpc-users.org...
    > On Mon, 12 Mar 2007 16:36:48 -0500, JohnQ wrote:
    >
    >> having a neutral foundation at the kernal level upon
    >> which to build whatever you want without having to ascribe to a certain
    >> model. If it's too UNIX-y, I'm not going to like it.

    >
    > From memory of reading the code pre 1.5-era, Minix is pretty Unix-y all
    > the way down. (Unix V7!) All you've got is processes (CSP-style). If
    > you want anything that speaks threads or even lower-level, then you'll
    > need to start from Mach or L4 (or something else).
    >
    > What have you got against posix? It's simple and it works.


    I find it cryptic rather than simple. I like the Windows event, threading
    and process models.

    > You could look into the iTron flavour for eCOS, if you're after something
    > different. Don't know how up-to-date that is, though. I think that the
    > posix API gets most of the running.


    I was just wondering if Minix in the future was going to be something I'd
    want to build on top of (GUI, mem mgmt, stuff like that). Apparently, if
    within every main() function there are signals being caught, forks etc., I
    think that's a bit long in the tooth already.

    John



  13. Re: Minix = Just another Free UNIX?


    "JohnQ" writes:

    > want to build on top of (GUI, mem mgmt, stuff like that). Apparently, if
    > within every main() function there are signals being caught, forks etc., I
    > think that's a bit long in the tooth already.


    What are you talking about? I'm suddenly ridden by the suspicion that
    you have hardly ever programmed C (neither under Unix nor under
    Windows). "within every main() function"? What? If that is so your
    plans (to reform the operating systems world) are somewhat at odds
    with your experience.

    And BTW: You didn't answer my questions about your alernative API.

    Regards -- Markus


+ Reply to Thread