Text 1709, 96 rader
Skriven 2004-11-01 20:01:55 av Michiel Broek (2:280/2802)
   Kommentar till text 1706 av Francois Thunus (2:270/25.2)
Ärende: Re: HPPA ?
==================
Hi Francois,

Francois Thunus wrote to Michiel Broek:
FT> * Replying to a msg in LINUX_BBS (Fido  : LinuxBBS )
FT> 
FT> Hello Michiel.
FT> 
FT> 23 Oct 04 13:16, Michiel Broek wrote to Francois Thunus:
FT> 
FT>  FT>> At this stage I am trying to extract the few files I'll need by
FT>  FT>> studying the makefile. then I can transfer the files to the arm
FT>  FT>> machine, compile ar, and finally compile gzip and bzip2 there
FT>  FT>> (hopefully). Once I have that, I'll start building the rest I need,
FT>  FT>> like the unix interactive tools, (mc is way too big for what I
FT>  FT>> need), etc. I'll probably recompile busybox too, because the version
FT>  FT>> I have is still a pre 1.0 (RC4 I think).
FT> 
FT>  MB> I would not do that, compile the whole package because it also 
FT> contains
FT>  MB> the other important tools to create the final binaries.
FT> 
FT> two things:
FT> 1) the package is big, and I'm notsure I have room enough

I know and you also need a lot of temp space during the compile.

FT> 2) I am supposed to have most of those already. The problem i'm running 
FT> into is that the "ar" provided by busybox does not offer some of the 
FT> functionalities needed by my makefile. I'm going to need a (f)lex too, but 

FT> for the rest, I shouldbe ok with what was provided by the zgcc package. 
FT> (at least until proven otherwise...)

Flex is not big. For the big packages, I mount a temp space from another
machine via nfs, to compile glibc I need over 1 Gig of free diskspace. I
don't know if you can do the same.

FT> 
FT>  FT>> Given time and disk space, I might even try to compile mbbs just for
FT>  FT>> the fun of it :-) I'm using zgcc 2.95.
FT> 
FT>  MB> Most of the stuff for big-endian machines is ok. I guess the arm is
FT>  MB> bigendian too. A few months ago someone noticed that the jam 
FT> mesagebase
FT>  MB> files are on a bigendian machine are written in bigendian format 
FT> while
FT>  MB> all pointers must be in intel format. This just means you cannot 
FT> share
FT>  MB> the message bases via nfs with little endian machines. I will fix 
FT> that
FT>  MB> someday but doing so needs also a tool that "repairs" the existing
FT>  MB> messsage bases. On the machine itself it works of course.
FT> Such a tools shouldn't be too difficult to write I guess. just a matter of 

FT> swapping a couple of bytes in a binary files :-)

Indeed, that shouldn't be too hard to do.

FT> 
FT>  FT>> It is fun to play with a different platform than intel for a change.
FT>  FT>> I'm back in the old days of wondering, pondering and hacking my way
FT>  FT>> thru sources :-)
FT> 
FT>  MB> It sure is. Nothing is the same and comes natural. But it's a great
FT>  MB> learning experience. I have here the hp and an old Sparcstation-2. 
FT> Both
FT>  MB> slow but they are reliable and do their job.
FT> 
FT> Sparc is cool. I work with SunOs for a while. Mind you, if you install 
FT> gcc, half of your troubles are already gone. I was working with the 
FT> default compiler for a while, and switching to gcc was a real 
FT> eye-opener...

The Sparc runs NetBSD, it's an old machine and Solaris 2.6 was too much for
that machine. With NetBSD gcc is available.

With Solaris, all tools we are "used to have" are available, that's the good
part to make Solaris more comfortable.


FT> Enjoy.
FT> Will download 0.7 while I'm at it :-)

Oke.



    Greetings, Michiel Broek

Email:   mbse@mbse.dds.nl
Fidonet: Michiel Broek at 2:280/2802

... Laws are like bones; they're made to be broken.

--- MBSE BBS v0.70.0 (GNU/Linux-i386)
 * Origin: MBSE Linux BBS. Made in the Netherlands (2:280/2802)