On Thu, 30 Dec 2004 22:15:19 -0000, jms1 <[EMAIL PROTECTED]> wrote:
Dear Peter
How do you get on the QL developers list? I tried sometime ago and never heard a thing
John
Its the link at the very bottom of this email to subscribe - ql-developers was moved across to q-v-d.com as well as ql-users (hence your cc is wrong)
http://www.q-v-d.demon.co.uk/smsqe.htm
is the info page
----- Original Message ----- From: <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Cc: <ql-developers@quanta.org.uk> Sent: Tuesday, November 16, 2004 1:28 PM Subject: Re: [ql-users] qdos gcc problem
step-Michael Grunditz wrote:
> The qdos-gcc libc does not compile on ppc linux, any hints ?
You could try to post such issues to the ql-developers list. Maybe some who deal with crosscompiling or Linux are still subscribed there. I only have a sloppy look at ql-users archives now and then, but that happens rather seldom. So if you want me to see follow-ups soon, please keep the cc ql-developers.
Before going into the details, it needs to be checked which exact GCC sources you are using, and which patches you have applied. There is aby-step instruction on how to build qdos-gcc at
http://thgodef.nerim.net/qdos/qdosgcc.html
Do you refer to that?
If so, I think it can't work on PPC Linux without further modification. I
have not specificly dealt with PPC Linux, but built qdos-gcc for 68k
Linux, and had several endianess related problems. You probably know that
x86 stores data in little-endian format, but the PPC (although it could
hardware-wise also run in little-endian mode) uses big endian like 68k.
There's a way to fix the endianess problems, but I don't remember the
exact steps from memory. I could reproduce and document them for you with
some effort, but I have little motivation to do so, for another reason:
I found that qdos-gcc generates wrong code from long macros, which are
e.g. used in some usual string operations. I reproduced the same effect
under x86 Linux, 68k Linux and cygwin, so I'm pretty sure it's not an OS
or library issue. I created a simplified example which eases reproduction
of this problem, but nobody ever seemed (able) to care. Consequently, I
stopped using qdos-gcc, it's too unreliable for me. The QL has no source-
level C debugger, so it is already hard enough to debug larger projects
_without_ additional serious compiler problems.
In the end I had to turn back to good old native C68. Recently I have also
patched the C68 (+assembler +linker +archiver) sources for MacOS X. The
resulting cross-toolchain (PPC/MacOSX to 68k/QDOS) seemed to work OK. At
least I was able to build my ethernet, tcp, email and web stuff, including
the assembler sources. I had no time for further tests yet.
If it helps, and somebody has the webspace, I can upload my C68 stuff for
MacOS X somewhere. That would be "quick and dirty" as it is. I don't know
exactly how much MacOS X and Linux differ with respect to compiling C68,
but since both are POSIX compliant and use GCC, there shouldn't be too
much.
An alternative might be to hunt for Jonathan Hudson's old XTC68 sources.
BTW it's a pity that the latest C68 sources are not accessible. Classic C68 has quite oldish behaviour, e.g. it doesn't even tolerate // comment markers. Some time ago I notified Dave about the problems with his website, but they don't seem fixed yet.
All the best Peter
_______________________________________________ QL-Users Mailing List http://www.quanta.org.uk/mailing.htm
_______________________________________________ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
-- Rich Mellor RWAP Services 26 Oak Road, Shelfield, Walsall, West Midlands WS4 1RQ
http://www.rwapservices.co.uk/
_______________________________________________ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
