Create an X Workstation - Xwindows

This is a discussion on Create an X Workstation - Xwindows ; Hello, Does anyone here know about a good howto on creating an own X11 distribution? I'm trying to compile XFree86 3.3.6 (as this is the last one supporting my WD90C24 chipset), but it doesn't compile through. Furthermore, when I get ...

+ Reply to Thread
Results 1 to 12 of 12

Thread: Create an X Workstation

  1. Create an X Workstation

    Hello,

    Does anyone here know about a good howto on creating an own X11
    distribution?

    I'm trying to compile XFree86 3.3.6 (as this is the last one supporting my
    WD90C24 chipset), but it doesn't compile through. Furthermore, when I get
    the XF86_SVGA server compiled, it doesn't want to start. Without any error
    message, I'm getting "Fatal server error: no screens found". A binary from
    somewhere starts without any problem, but takes quite a bit more money (as
    this is a 386-based-low-mem workstation, this is a problem).

    Anyway, does somebody have some howto about compiling X11R6/SVGA and running
    it? What in particular is needed in which locations on the disk? These are
    my questions.

    Any help is highly appreciated.

    Regards,
    Sebastian



  2. Re: Create an X Workstation

    Hello,

    > You haven't shown us any of the commands you used.
    > You also haven't posted any of the compile errors.


    I will write exactly what I've done:

    I got X336src-1.tgz and X336src-2.tgz and unpacked them.
    The source directory is /home/basti/linux/work/xfree/xc.

    gcc is a symlink to i486-unknown-linux-gnu-gcc,
    same goes for ar, as, c++, ld, ranlib, strip and cc.

    In xc/programs/Xserver/hw/xfree86/vga256:
    I ran "./cvga256.sh generic pvga1" which creates a vga256Conf.c containing
    only references to drivers I need (wd90c24 is inside the driver pvga1) after
    doing "chmod +x cvga256.sh".

    In xc/programs/Xserver/hw/xfree86/os-support/linux/lnx_init.c I changed
    "void xf86OpenConsole" to reflect devfs-names (I replaced "/dev/tty" with
    "/dev/vc/" except in line 152, which opens /dev/tty).

    In xc/config/cf/site.def I defined HasGcc2 (I use GCC 3.4.3 in my
    toolchain), then I defined ProjectRoot to be /usr/X11 (I want to X to sit
    there).

    Then I ran "make World 2>&1 | tee world.log" in the xc-directory.

    Some programs weren't compiled because of errors, e.g. twm or xclock.
    Nevertheless the compiling process won't stop. After about 32 minutes the
    compiling is done: "Full build of Release 6.3 of the X Window System
    complete." But no servers are built.

    If I manually run "make 2>&1 | more" inside xc/programs/Xservers I get lots
    of errors like

    WaitFor.c: In function `WaitForSomething':
    WaitFor.c:187: error: structure has no member named `fds_bits'
    ....
    WaitFor.c:480: error: structure has no member named `fds_bits'
    make[1]: *** [WaitFor.o] Error 1
    make[1]: Leaving directory
    `/home/basti/linux/work/xfree/xc/programs/Xserver/os'
    make: *** [os] Error 2

    I don't know why this is happening. Can you point me somewhere?
    Once I was further, but I don't know what I did change.

    Regards,
    Sebastian



  3. Re: Create an X Workstation

    "Sebastian" writes:

    > Hello,
    >
    >> You haven't shown us any of the commands you used.
    >> You also haven't posted any of the compile errors.

    >
    > I will write exactly what I've done:
    >
    > I got X336src-1.tgz and X336src-2.tgz and unpacked them.
    > The source directory is /home/basti/linux/work/xfree/xc.
    >
    > gcc is a symlink to i486-unknown-linux-gnu-gcc,
    > same goes for ar, as, c++, ld, ranlib, strip and cc.
    >
    > In xc/programs/Xserver/hw/xfree86/vga256:
    > I ran "./cvga256.sh generic pvga1" which creates a vga256Conf.c containing
    > only references to drivers I need (wd90c24 is inside the driver pvga1) after
    > doing "chmod +x cvga256.sh".
    >
    > In xc/programs/Xserver/hw/xfree86/os-support/linux/lnx_init.c I changed
    > "void xf86OpenConsole" to reflect devfs-names (I replaced "/dev/tty" with
    > "/dev/vc/" except in line 152, which opens /dev/tty).
    >
    > In xc/config/cf/site.def I defined HasGcc2 (I use GCC 3.4.3 in my
    > toolchain), then I defined ProjectRoot to be /usr/X11 (I want to X to sit
    > there).
    >
    > Then I ran "make World 2>&1 | tee world.log" in the xc-directory.
    >
    > Some programs weren't compiled because of errors, e.g. twm or xclock.
    > Nevertheless the compiling process won't stop. After about 32 minutes the
    > compiling is done: "Full build of Release 6.3 of the X Window System
    > complete." But no servers are built.
    >
    > If I manually run "make 2>&1 | more" inside xc/programs/Xservers I get lots
    > of errors like
    >
    > WaitFor.c: In function `WaitForSomething':
    > WaitFor.c:187: error: structure has no member named `fds_bits'


    I get a lot of Google hits on that message, here's one:

    http://groups.google.com/group/linux...633ee71a194afc

    I think you need some source code updates.

  4. Re: Create an X Workstation

    Hello,

    > > WaitFor.c: In function `WaitForSomething':
    > > WaitFor.c:187: error: structure has no member named `fds_bits'

    >
    > I get a lot of Google hits on that message, here's one:
    >
    >

    http://groups.google.com/group/linux...hread/thread/7
    b35c1b4a6651509/31633ee71a194afc%2331633ee71a194afc

    Yes, but none from these results (as far as I looked through them) deals
    with XFree86-servers.
    Anyway, my problem isn't directly related to Debian.

    > I think you need some source code updates.


    Do you know where to get them? I am not able to program C, so I cannot do
    this myself. I don't even understand where this error comes from, because
    the lines don't contain "fds_bits" somewhere...

    As I already wrote, I am stuck with XFree 3.3.6.

    Regards,
    Sebastian



  5. Re: Create an X Workstation

    "Sebastian" writes:

    > Hello,
    >
    >> > WaitFor.c: In function `WaitForSomething':
    >> > WaitFor.c:187: error: structure has no member named `fds_bits'

    >>
    >> I get a lot of Google hits on that message, here's one:
    >>
    >>

    > http://groups.google.com/group/linux...hread/thread/7
    > b35c1b4a6651509/31633ee71a194afc%2331633ee71a194afc
    >
    > Yes, but none from these results (as far as I looked through them) deals
    > with XFree86-servers.


    If you look at the link I provided, you'll see that the error message
    in VNC is in the file WaitFor.c, the same filename you report an error
    in.

    My guess is that the WaitFor.c file was copied from XFree into VNC.

    If you can find the patch to VNC, it might apply to XFree or show
    what has to be done.

    > Anyway, my problem isn't directly related to Debian.


    Debian uses the same X Server as other distros.

    >> I think you need some source code updates.

    >
    > Do you know where to get them? I am not able to program C, so I cannot do
    > this myself. I don't even understand where this error comes from, because
    > the lines don't contain "fds_bits" somewhere...


    If you're not a programmer, you may be completely out of luck unless
    you want to pay someone.

    Most likely the fds_bits reference is thru some MACRO, like this one:

    /usr/include/X11/Xpoll.h:
    #define FD_SET(n, p) ((p)->fds_bits[(n)/NFDBITS

    > As I already wrote, I am stuck with XFree 3.3.6.


    You might want to do more research on that.
    I tried but didn't get far.
    I don't know why anyone would remove support for old hardware.

    -or-

    Maybe you want to replace the hardware?

    -or-

    You might have better results compiling 3.3.6 using the OS and tool
    chain from the 3.3.6 era. Personally, I wouldn't go there.
    Some of the Google hits seemed to indicate the version of gcc was
    part of the problem.

  6. Re: Create an X Workstation

    Hello,

    > >> > WaitFor.c: In function `WaitForSomething':
    > >> > WaitFor.c:187: error: structure has no member named `fds_bits'
    > >>
    > >> I get a lot of Google hits on that message, here's one:
    > >>

    > >

    http://groups.google.com/group/linux...hread/thread/7
    > > b35c1b4a6651509/31633ee71a194afc%2331633ee71a194afc
    > >
    > > Yes, but none from these results (as far as I looked through them) deals
    > > with XFree86-servers.

    >
    > If you look at the link I provided, you'll see that the error message
    > in VNC is in the file WaitFor.c, the same filename you report an error
    > in.


    Yes, you are right.

    > My guess is that the WaitFor.c file was copied from XFree into VNC.


    That might be, I didn't think about that.

    > If you can find the patch to VNC, it might apply to XFree or show
    > what has to be done.


    The patch itself does not modify WaitFor.c , but patches some other files
    which I couldn't find in my source tree. Only the configuration is changed a
    bit, I'll try that soon.

    > > Anyway, my problem isn't directly related to Debian.

    >
    > Debian uses the same X Server as other distros.


    Yes.

    > >> I think you need some source code updates.

    > >
    > > Do you know where to get them? I am not able to program C, so I cannot

    do
    > > this myself. I don't even understand where this error comes from,

    because
    > > the lines don't contain "fds_bits" somewhere...

    >
    > If you're not a programmer, you may be completely out of luck unless
    > you want to pay someone.
    >

    That wouldn't be nice :-(

    > Most likely the fds_bits reference is thru some MACRO, like this one:
    >
    > /usr/include/X11/Xpoll.h:
    > #define FD_SET(n, p) ((p)->fds_bits[(n)/NFDBITS
    >

    I don't understand that, sorry.

    > > As I already wrote, I am stuck with XFree 3.3.6.

    >
    > You might want to do more research on that.
    > I tried but didn't get far.


    Neither did I.

    > I don't know why anyone would remove support for old hardware.
    >

    The drivers weren't needed when XFree86 evolved to 4.0.0, so they weren't
    ported in first place. Later Changelogs do not mention them, so I believe
    they were never ported.

    > -or-
    >
    > Maybe you want to replace the hardware?
    >

    No. It is a hard-soldered graphics chip (notebook).

    > -or-
    >
    > You might have better results compiling 3.3.6 using the OS and tool
    > chain from the 3.3.6 era. Personally, I wouldn't go there.


    That is my problem - I am not skilled enough to build a complete toolchain
    myself, but I used some scripts, which worked. But they won't create any
    toolchain from that era. But I don't want to be in the very old ages - only
    XFree must be old, everything else should be quite modern.

    > Some of the Google hits seemed to indicate the version of gcc was
    > part of the problem.


    Yes. I cannot understand why the compiler version is that important in
    building Linux and its applications. I mean, old source didn't change over
    the years, but modern compilers don't get it right. A modern gcc cannot
    compile linux-0.99p2 or whatever "out of the box". Minix' "cc" does the job,
    but cannot compile linux-2.4.
    In my opinion, something is going wrong. But others have to make decisions,
    I'm no developer.

    Regards,
    Sebastian



  7. Re: Create an X Workstation

    "Sebastian" writes:

    > Hello,
    >
    >> >> > WaitFor.c: In function `WaitForSomething':
    >> >> > WaitFor.c:187: error: structure has no member named `fds_bits'
    >> >>
    >> >> I get a lot of Google hits on that message, here's one:
    >> >>
    >> >

    > http://groups.google.com/group/linux...hread/thread/7
    >> > b35c1b4a6651509/31633ee71a194afc%2331633ee71a194afc
    >> >
    >> > Yes, but none from these results (as far as I looked through them) deals
    >> > with XFree86-servers.

    >>
    >> If you look at the link I provided, you'll see that the error message
    >> in VNC is in the file WaitFor.c, the same filename you report an error
    >> in.

    >
    > Yes, you are right.
    >
    >> My guess is that the WaitFor.c file was copied from XFree into VNC.

    >
    > That might be, I didn't think about that.
    >
    >> If you can find the patch to VNC, it might apply to XFree or show
    >> what has to be done.

    >
    > The patch itself does not modify WaitFor.c , but patches some other files
    > which I couldn't find in my source tree. Only the configuration is changed a
    > bit, I'll try that soon.
    >
    >> Most likely the fds_bits reference is thru some MACRO, like this one:
    >>
    >> /usr/include/X11/Xpoll.h:
    >> #define FD_SET(n, p) ((p)->fds_bits[(n)/NFDBITS
    >>

    > I don't understand that, sorry.


    You say you're not a programmer, but here goes:

    Given the above "#define", if the program
    contained:

    FD_SET(4,fd_struct_ptr)

    The #define would change that to:

    ((fd_struct_ptr)->fds_bits[(4)]/NFDBITS

    (The "n" and "p" in FD_SET get substituted into
    the code on the right.)

    The code shown above says that fd_struct_ptr contains (->)
    the field fds_bits.

    Somewhere else, the type of fd_struct_ptr is declared,
    probably like this:

    fd_set *fd_struct_ptr;

    The error says that fds_bits is not part of a structure of type
    fd_set.

    Sufficiently confused?

    The code you are trying to compile contains something
    that causes that error, maybe a missing #include
    or a wrong #include. To use FD_SET, you need:

    #include

    Many other things could be wrong.

    The macro might not be FD_SET.
    There is a compile flag (-E) that you can
    use that runs the .c file thru just the preprocessor
    and creates tbe pre-processed file on stdout.
    In stdout you will be able to see exactly what line
    does the reference to fds_bits.

    >> Maybe you want to replace the hardware?
    >>

    > No. It is a hard-soldered graphics chip (notebook).


    Bummer.

    >> -or-
    >>
    >> You might have better results compiling 3.3.6 using the OS and tool
    >> chain from the 3.3.6 era. Personally, I wouldn't go there.

    >
    > That is my problem - I am not skilled enough to build a complete toolchain
    > myself, but I used some scripts, which worked. But they won't create any
    > toolchain from that era. But I don't want to be in the very old ages - only
    > XFree must be old, everything else should be quite modern.
    >
    >> Some of the Google hits seemed to indicate the version of gcc was
    >> part of the problem.

    >
    > Yes. I cannot understand why the compiler version is that important in
    > building Linux and its applications. I mean, old source didn't change over
    > the years, but modern compilers don't get it right. A modern gcc cannot
    > compile linux-0.99p2 or whatever "out of the box". Minix' "cc" does the job,
    > but cannot compile linux-2.4.
    > In my opinion, something is going wrong. But others have to make decisions,
    > I'm no developer.


    I don't think it's gcc by itself, but gcc includes the headers in /usr/include.

    I was thinking you could install some distro contemporary with 3.3.6
    like Mandrake 8.1. I think the ISOs can still be found.

  8. Re: Create an X Workstation

    Hello,

    > >> > WaitFor.c: In function `WaitForSomething':
    > >> > WaitFor.c:187: error: structure has no member named `fds_bits'


    [...]
    > My guess is that the WaitFor.c file was copied from XFree into VNC.


    I think so, too, after looking at the file and the patch given in the link.

    > If you can find the patch to VNC, it might apply to XFree or show
    > what has to be done.


    I patched the files manually, the line numbers were identical.
    But it didn't help, the error message is the same. In my opinion this patch
    helps at some point, but no changes were made which contained something like
    "fds_bits". Only some mallocs() were removed...

    I'm beginning to believe I will never be able to get it to work ... it's a
    pity.
    This is why I dislike Linux a bit (although I use it up to some point) - it
    simply doesn't work without failures.

    Anyway, thank you for your help. I give up.

    Regards,
    Sebastian



  9. Re: Create an X Workstation

    "Sebastian" writes:

    > Hello,
    >
    >> >> > WaitFor.c: In function `WaitForSomething':
    >> >> > WaitFor.c:187: error: structure has no member named `fds_bits'

    >
    > [...]
    >> My guess is that the WaitFor.c file was copied from XFree into VNC.

    >
    > I think so, too, after looking at the file and the patch given in the link.
    >
    >> If you can find the patch to VNC, it might apply to XFree or show
    >> what has to be done.

    >
    > I patched the files manually, the line numbers were identical.
    > But it didn't help, the error message is the same. In my opinion this patch
    > helps at some point, but no changes were made which contained something like
    > "fds_bits". Only some mallocs() were removed...


    Those were probably malloc prototypes something like:

    void *malloc(size_t size);

    on a line by itself?

    If so, they should be replaced by:

    #include

    at the front of the file.

    I never looked closely but that's what I saw.
    It seemed odd that they were messing with malloc to solve
    this problem. I must have missed something.

    > I'm beginning to believe I will never be able to get it to work ... it's a
    > pity.
    > This is why I dislike Linux a bit (although I use it up to some point) - it
    > simply doesn't work without failures.


    Of course this has NOTHING to do with Linux.

    It doesn't even have anything to do with a particular distribution.
    You've struck out on your own trying to piece something together.

    > Anyway, thank you for your help.


    You're welcom.

    > I give up.


    Can't fault you there.
    Just trying to build xc is a non-trivial undertaking.

  10. Re: Create an X Workstation

    Hello,

    [...]
    > > I patched the files manually, the line numbers were identical.
    > > But it didn't help, the error message is the same. In my opinion this

    patch
    > > helps at some point, but no changes were made which contained something

    like
    > > "fds_bits". Only some mallocs() were removed...

    >
    > Those were probably malloc prototypes something like:
    >
    > void *malloc(size_t size);
    >
    > on a line by itself?
    >

    I have a general understanding of programming (Delphi at school and some
    dialects of Basic), but this code is a bit too much for me. Here the lines I
    changed (as far as I could read the diff):

    file 1: (AuRead.c)
    - char *data, *malloc ();
    + char *data;
    - char *malloc ();
    file 3: (AuWrite.c)
    - char *malloc ();

    I have no idea what is done in these files.

    > If so, they should be replaced by:
    >
    > #include
    >
    > at the front of the file.
    >

    I'm not sure if that's the cure.

    > I never looked closely but that's what I saw.
    > It seemed odd that they were messing with malloc to solve
    > this problem. I must have missed something.
    >

    Same as me. Afaik malloc() deals with memory reservation (and freeing), but
    has nothing to do with some error conditions...

    > > I'm beginning to believe I will never be able to get it to work ... it's

    a
    > > pity.
    > > This is why I dislike Linux a bit (although I use it up to some point) -

    it
    > > simply doesn't work without failures.

    >
    > Of course this has NOTHING to do with Linux.
    >

    X11 is a primary part of any home user linux distribution. I don't know
    about any other supported graphics subsystem except XFree86 / Xorg.
    SVGALib is an idea, or directly the framebuffer device. But most software
    simply doesn't work with these subsystems. And they need Vesa support, which
    my chipset doesn't have.

    In my opinion it has something to do with Linux, as there are many pieces,
    which sometimes work well, sometimes don't work. Many programs exist for any
    purpose, but there is no "general standard". Every X program has its own
    look'n'feel, QT applications look and behave different than GTK
    applications, nothing is standardized. I don't know why, but this is a
    reason to stay away. Even if many people don't like this attitude...

    I will continue using Linux, no question, but as far as I can tell, it is
    not good enough yet to be an (my) all-purpose primary operating system.
    Sorry.

    > It doesn't even have anything to do with a particular distribution.
    > You've struck out on your own trying to piece something together.
    >

    Yes. It works somehow, but without any graphics.
    Midnight Commander works (although without gpm), which isn't a trivial task
    either, same goes for pcmcia-cs (Ethernet). I even got SVGATextMode running.

    Why has X11 to be so complicated? There are programs, far more complex,
    which simply "compile" and "work". Most software configures for
    cross-compiling - X doesn't without tricks and lots of work.

    > > Anyway, thank you for your help.

    >
    > You're welcome.
    >
    > > I give up.

    >
    > Can't fault you there.
    > Just trying to build xc is a non-trivial undertaking.


    Yes, but why?
    Shouldn't it simply build?
    Shouldn't it simply work?
    Why not?

    Regards,
    Sebastian



  11. Re: Create an X Workstation

    "Sebastian" writes:

    > Hello,
    >
    > [...]
    >> > I patched the files manually, the line numbers were identical.
    >> > But it didn't help, the error message is the same. In my opinion this

    > patch
    >> > helps at some point, but no changes were made which contained something

    > like
    >> > "fds_bits". Only some mallocs() were removed...

    >>
    >> Those were probably malloc prototypes something like:
    >>
    >> void *malloc(size_t size);
    >>
    >> on a line by itself?
    >>

    > I have a general understanding of programming (Delphi at school and some
    > dialects of Basic), but this code is a bit too much for me. Here the lines I
    > changed (as far as I could read the diff):
    >
    > file 1: (AuRead.c)
    > - char *data, *malloc ();
    > + char *data;
    > - char *malloc ();
    > file 3: (AuWrite.c)
    > - char *malloc ();
    >
    > I have no idea what is done in these files.


    Yes, what you have there is a 'prototype' for malloc.
    It's just a statement that malloc returns a character pointer.
    Lots of older programs contain prototypes like that, but they should not.

    The 'correct' prototype for malloc is in stdlib.h.

    >> If so, they should be replaced by:
    >>
    >> #include
    >>
    >> at the front of the file.
    >>

    > I'm not sure if that's the cure.
    >
    >> I never looked closely but that's what I saw.
    >> It seemed odd that they were messing with malloc to solve
    >> this problem. I must have missed something.
    >>

    > Same as me. Afaik malloc() deals with memory reservation (and freeing), but
    > has nothing to do with some error conditions...
    >
    >> > I'm beginning to believe I will never be able to get it to work ... it's
    >> > a
    >> > pity.
    >> > This is why I dislike Linux a bit (although I use it up to some point) -
    >> > it
    >> > simply doesn't work without failures.

    >>
    >> Of course this has NOTHING to do with Linux.
    >>

    > X11 is a primary part of any home user linux distribution. I don't know
    > about any other supported graphics subsystem except XFree86 / Xorg.
    > SVGALib is an idea, or directly the framebuffer device. But most software
    > simply doesn't work with these subsystems. And they need Vesa support, which
    > my chipset doesn't have.
    >
    > In my opinion it has something to do with Linux, as there are many pieces,
    > which sometimes work well, sometimes don't work. Many programs exist for any
    > purpose, but there is no "general standard". Every X program has its own
    > look'n'feel, QT applications look and behave different than GTK
    > applications, nothing is standardized. I don't know why, but this is a
    > reason to stay away. Even if many people don't like this attitude...


    Not all programs look or act the same on Windows either.
    (From the little I've seen.)

    From what I hear, Mandriva has themed QT and GTK so that they look and
    act pretty much the same.

    Personally, I'm not too interested, I use Linux because I want
    total control of every aspect of my computer.
    KDE and Gnome aren't appealing to me.
    I want something I can completely control (fvwm2).

    > I will continue using Linux, no question, but as far as I can tell, it is
    > not good enough yet to be an (my) all-purpose primary operating system.
    > Sorry.


    Well, it's all semantics.
    "Linux" often refers to just the kernel.
    The way you are trying to use the term
    I think you are referring to the distribution.
    But you aren't using a particular distribution.
    You're mixing together parts in a way that no developer
    expected you to.

    >> It doesn't even have anything to do with a particular distribution.
    >> You've struck out on your own trying to piece something together.
    >>

    > Yes. It works somehow, but without any graphics.
    > Midnight Commander works (although without gpm), which isn't a trivial task
    > either, same goes for pcmcia-cs (Ethernet). I even got SVGATextMode running.
    >
    > Why has X11 to be so complicated? There are programs, far more complex,
    > which simply "compile" and "work". Most software configures for
    > cross-compiling - X doesn't without tricks and lots of work.
    >
    >> > Anyway, thank you for your help.

    >>
    >> You're welcome.
    >>
    >> > I give up.

    >>
    >> Can't fault you there.
    >> Just trying to build xc is a non-trivial undertaking.

    >
    > Yes, but why?
    > Shouldn't it simply build?
    > Shouldn't it simply work?
    > Why not?


    X (and XFree) have a long history.
    For some reason, X developed it's own build system (Imake).
    The first time I had to build X, I was reading docs and
    scratching my head for a long time.

    The good news is that recent releases of Xorg let you do
    the regular configure/make routine.

    You should understand that by using XFree 3.3.6, you are
    subjecting yourself to the state of XFree from quite a few
    years ago.

    By compiling from source, you are opening a can of worms that
    is normally handled by the distro creator.

    I still think you could shortcut this whole mess by locating
    the ISOs for Mandrake 8.1 and doing a straight install.
    I believe that used XFree 3.3.6 or pretty close to it.

  12. Re: Create an X Workstation

    Hello,

    [...snip...]

    > Not all programs look or act the same on Windows either.
    > (From the little I've seen.)


    Most do.

    > From what I hear, Mandriva has themed QT and GTK so that they look and
    > act pretty much the same.


    Yes, the look might be the same, but I cannot believe they act the same way.

    > Personally, I'm not too interested, I use Linux because I want
    > total control of every aspect of my computer.
    > KDE and Gnome aren't appealing to me.
    > I want something I can completely control (fvwm2).


    I prefer IceWM, but this is a matter of personal likings.

    > > I will continue using Linux, no question, but as far as I can tell, it

    is
    > > not good enough yet to be an (my) all-purpose primary operating system.
    > > Sorry.

    >
    > Well, it's all semantics.
    > "Linux" often refers to just the kernel.

    Yes. The rest is build by GNU and other software projects.

    > The way you are trying to use the term
    > I think you are referring to the distribution.


    My own distribution, which doesn't exist :-)

    > But you aren't using a particular distribution.


    I wanted to create one myself.

    > You're mixing together parts in a way that no developer
    > expected you to.


    I tried to to what every distributor does.

    > X (and XFree) have a long history.
    > For some reason, X developed it's own build system (Imake).
    > The first time I had to build X, I was reading docs and
    > scratching my head for a long time.


    Same did I, but I couldn't even get out of the docs why it didn't work.

    > The good news is that recent releases of Xorg let you do
    > the regular configure/make routine.


    Wow, which invention...

    > You should understand that by using XFree 3.3.6, you are
    > subjecting yourself to the state of XFree from quite a few
    > years ago.


    The drivers were never ported, so the chipset doesn't work with later
    revisions.

    > By compiling from source, you are opening a can of worms that
    > is normally handled by the distro creator.


    Yes. But I don't know why these "worms" exist.
    In my opinion something should be done in a way which lets you use these
    files even some years later. I don't want to be forced to install any
    mid-80's system only to be able to read those old floppies lying around.
    Support which once did exist should be kept, at least partially. But this is
    my opinion and I know it is not easy or applicable everywhere.

    > I still think you could shortcut this whole mess by locating
    > the ISOs for Mandrake 8.1 and doing a straight install.
    > I believe that used XFree 3.3.6 or pretty close to it.


    It doesn't work. Neither did Debian 3.0.
    I cannot accept a booting time of 20 minutes (Debian 3.0) or 12 minutes
    (Debian 2.0), sorry.

    I lay down the work and try to find something better - or let it be and use
    Windows. This works at least.

    Regards,
    Sebastian



+ Reply to Thread