jkh@meepmeep.pcs.com (Jordan K. Hubbard) (04/06/91)
Some said it couldn't be done. Others said it shouldn't be done. Still
more said "Just what the hell is a Symmetric, anyway?"
But it was done.
GCC 1.39 and GAS 1.38 have been ported to the Symmetric system and are
hereby offered free of charge to the other 3 Symmetric system owners
out there ("What? You turned yours into a planter?? Oh. Make that 2 other
owners..").
Many thanks go to Ian Dall for his ICM 32016 port of GCC 1.37, after
which this port is very closely modeled. Thanks must also go to the
Folgers coffee company, without whos product the project would never have
been completed.
The release is somewhat unusual as GNU software releases go, since I
won't actually be distributing that much in the way of diffs. This
is primarily because Symmetric's native C compiler was TOO BRAIN DAMAGED
to actually compile C on its own (I tried, believe me) and I was forced
to actually generate the bootstrap binary on a different machine by
cross-compiling. This bootstrapped copy has since stage 3 compiled itself
on the Symmetric and will be distributed as a set of binary images.
Accompanying the binaries will be several configuration files for both GCC
and GAS, a few diffs and full instructions on obtaining the latest sources
from the FSF that you can then use to regenerate the entire environment at
will, if you so desire.
Since I personally loath making tapes (and people that have counted on
me for tapes in the past have since come to loath me as well), the one
Symmetric owner in Germany has consented to do the grunt work. He also
has an ulterior motive in wanting to get the names and addresses of those
other 2 symmetric owners (it gets lonely being an orphan sometimes), so
he doesn't mind.
To get your hands on a copy of GCC for the Symmetric (soon to be a
collectors item), send appropriate media (50 Mbyte TEAC/SCS Cassette only!)
plus return postage in any suitably convertable western currency to:
Julian Stacey
Holz Strasse 27d
D 8000 Munich 5
Germany.
Phone: + 49 89 268616 (Timezone=GMT+1 10:00-21:00 local time).
Media: 50 Mbyte TEAC/SCS Cassette (No Cartridges, 9 track, FTP).
E-mail: stacey@guug.de
If you happen to have any interesting Symmetric lore (other than
the fact that they went out of business - he knows that already) to
share, I'm sure he'd appreciate hearing from you as well.
If a need exists, and you ask me nicely, I'll also put the whole mess up
for anonymous ftp someplace and save you the cost of a mailing. I'm not
doing this at the outset simply because I don't know if this offer is
going to drop into the void or not and I don't feel like pushing the
bits across for nothing.
With scarcely concealed anticipation,
I remain,
Jordan Hubbardscott@freedm.UUCP (Scott Stratmoen) (04/07/91)
In article <m0jOsjJ-0003N0C@meepmeep.pcs.com> jkh@MEEPMEEP.PCS.COM (Jordan K. Hubbard) writes: >Some said it couldn't be done. Others said it shouldn't be done. Still >more said "Just what the hell is a Symmetric, anyway?" > >But it was done. Just for curiosity, does anyone still run a ICM-3216? If so, is there any interest in native coff GCC-1.39, G++-1.39 and GDB-1.35 for said platform? This tool set is intended for SystemVr2. No testing has been done on the ICM-3216 System Vr3.0 port. Note that these tools have been known to work on an ICM-332-1 running r3.0 Fine print: - The code and binaries are available, but I don't want to get into major amounts of tape dup'ing.... - In addition, as the native pcc is not able to compile gcc, you either need some other compiler (i.e. gnx, pcc from the ICM-332-1, etc) or the binary of gcc. - There still remains two known bugs, one caused by a md problem which appears in less than 1% of optimizations. The result is a compiler abort due to an impossible register allocation. The other is due to a problem in the native assembler. This is a minor problem involving typedef's of functions when debugging information is in the as source. -- Scott Stratmoen | ..!ast!freedm!scott | scott%freedm@ast.dsd.northrop.com
ian@sibyl.eleceng.ua.OZ (Ian Dall) (04/09/91)
In article <197@freedm.UUCP> scott@freedm.UUCP (Scott Stratmoen) writes: >Just for curiosity, does anyone still run a ICM-3216? I do, but maybe not for much longer. (Just as soon as I get an operating system running on my PC532). > If so, is there any >interest in native coff GCC-1.39, G++-1.39 and GDB-1.35 for said platform? ^^^^^^ As you note (later) there is an assembler bug which IMO makes native mode undesirable. > - In addition, as the native pcc is not able to compile gcc, you either > need some other compiler (i.e. gnx, pcc from the ICM-332-1, etc) or > the binary of gcc. Native (as in uses the native "as" and "ld") gcc binary's for the ICM 3216 are available for anonymous ftp from augean.eleceng.ua.oz. This is (I think) 1.37, but it is suitable for bootstrapping. > - There still remains two known bugs, one caused by a md problem which > appears in less than 1% of optimizations. Do you have code to exercise this bug? I would be interested in checking whether I have it or not in my port. > The other is > due to a problem in the native assembler. This is a minor problem > involving typedef's of functions when debugging information is > in the as source. There is also a bug in ld which manifests its self from time to time. I think it also has to do with debugging information (I knew, but I forget). It is problems like these which made me, eventually, go for the coff-encapsulation stuff so I could use gas and binutils. Ian Dall -- Ian Dall I'm not into isms, but hedonism is the most harmless I can think of. -- Phillip Adams ACSnet: ian@sibyl.eleceng.ua.oz internet: ian@sibyl.eleceng.ua.oz.au