Re: GNUMach 1.x fixed (autoconf+kb)

[Marco Gerards]

"Alfred M. Szmidt" <ams@kemisten.nu> writes:

>    I've included the regenerated autoconf output in this patch
>    (configure, etc.).
> 
> Eech, don't.  No need to bloat the patch with unneeded info that can
> be regenerated.

Yeah, that was stupid. :)

>    The queues of the keyboard driver were not initialized until the
>    device was opened, this caused a crash if a user pressed a key
>    while GNUMach was still booting.  I've fixed this.
> 
> I would suggest spliting it up, one kb patch, and one that touches
> autoconf stuff.

Why?  Those other changes are quite small.  And I just don't see the
point of splitting it up...

[...]

>    +This version has support for autoconf 2.57, gcc 3.4 and binutils
>    +2.14.x.
> 
> 3.4 is not released.

Ok, you fixed it to work with a gcc version because it was broken.
Was that gcc 3.3?  And it was possible that you fixed it to work with
the CVS version of gcc.

>    diff -upNr ../gnumach.orig/configure.in ./configure.in
>    --- ../gnumach.orig/configure.in   2004-01-19 19:30:37.000000000 +0100
>    +++ ./configure.in 2004-01-19 18:41:30.000000000 +0100
> 
> Maybe a rename is in order, newer autoconf's prefer to have
> configure.ac.

I agree.  The same must be done for oskit-mach and the Hurd (and
perhaps mig too).

>    @@ -28,11 +27,6 @@ i[[3456]]86) systype=i386 ;;
>     *) AC_MSG_ERROR([unsupported CPU type]) ;;
>     esac
> 
>    -case "$host_os" in
>    -gnu*) ;;
>    -*) AC_MSG_ERROR([sorry, this is the gnu os, not $host_os]) ;;
>    -esac
>    -
> 
> Why did you remove it?

Because I wanted to make the configure.in to work like the one of
oskit-mach.  And what is the reason of this code?  You can perfectly
build a GNUMach kernel using GNU/Linux for example.

Thanks,
Marco