[gnu.g++.help] g++ and libg++ on SCO Opendesktop

trevor@trevan.uucp (01/01/91)

I have been trying to get libg++ working on SCO Opendesktop.
When g++ has compiled the tests test0 core dumps. I would like to
know if anyone has these working on SCO. Here are the details

		gcc 1.37.1
		gas 1.36.0
		g++ 1.37.1
		libg++ 1.37.0

I have tried with and without the gnu-coff patches.
With the coff patches gcc works well with gas. I have tried g++
with and without using the ifile.
Make_ifile assums a value of 0x400000 if NBPS is not defined. Is this
correct for SCO.
-- 
				regards trevor
						trevor@trevan.uucp

mb@ttidca.TTI.COM (Michael Bloom) (01/03/91)

In article <1991Jan01.111531.15463@trevan.uucp> trevor@trevan.uucp writes:

>When g++ has compiled the tests test0 core dumps. I would like to

Test0 tries to perform incremental loading, using a feature of the
berkeley loader that is not present on system 5 systems.  If I get the
time, I might include a front end to ld that performs this relatively
trivial task in a future version of the coff set. Then again, with
some recent developments, that may turn out not to be necessary.

>With the coff patches gcc works well with gas. I have tried g++
>with and without using the ifile.
>Make_ifile assums a value of 0x400000 if NBPS is not defined. Is this
>correct for SCO.

I've been told that this value works for all 386 based system V systems.

On a separate matter, One thing you should take care to be certain
of is that libg++ contains the version of gnulib3 that is patched in
the g++ directory and that it is compiled with -DUSE_GPLUS_IFILE.

Not including an extra copy of the patch for use in the libg++
directory was an oversight on my part that has tripped up a number of
people (although some had it figured out by the time my email reply
had reached them). The change in the patched version is necessary in
order for the global constructors and destructors to get called.

james@bigtex.cactus.org (James Van Artsdalen) (01/04/91)

In <22319@ttidca.TTI.COM>, mb@ttidca.TTI.COM (Michael Bloom) wrote:

| Make_ifile assums a value of 0x400000 if NBPS is not defined. Is this
| correct for SCO.

> I've been told that this value works for all 386 based system V systems.

What value is this, the start of .data space?  If so, no, that value
is not correct for all 386 systems.  SysVr4 puts .data somewhere else.
-- 
James R. Van Artsdalen          james@bigtex.cactus.org   "Live Free or Die"
Dell Computer Co    9505 Arboretum Blvd Austin TX 78759         512-338-8789

rfg@NCD.COM (Ron Guilmette) (01/13/91)

In article <22319@ttidca.TTI.COM> mb@ttidca.TTI.COM (Michael Bloom) writes:
>In article <1991Jan01.111531.15463@trevan.uucp> trevor@trevan.uucp writes:
>
>>When g++ has compiled the tests test0 core dumps. I would like to
>
>Test0 tries to perform incremental loading, using a feature of the
>berkeley loader that is not present on system 5 systems.

That statement should be ammended to say "... on PRE-SVR4 System 5 systems."
In SVR4 you get dynamic loading.

>If I get the
>time, I might include a front end to ld that performs this relatively
>trivial task in a future version of the coff set. Then again, with
>some recent developments, that may turn out not to be necessary.

I'm not 100% sure what "recent developments" Michael Bloom is refering
to, but perhaps he is talking about System 5 Release 4.  Considering
hom many improvements V.4 has over older System V releases, I certainly
hope and pray that people will not spend an inordinate amount of time
(from now on) trying to make GNU tools do clever things (e.g. dynamic
loading) on older System V releases.  All such work will be man-hours
down the drain before too long.

-- 

// Ron Guilmette  -  C++ Entomologist
// Internet: rfg@ncd.com      uucp: ...uunet!lupine!rfg
// Motto:  If it sticks, force it.  If it breaks, it needed replacing anyway.