[csw-maintainers] gcc4

[Mike Watters]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I thought I had the merge all worked out.
I was testing the "merged" binaries using the standard gcc testsuite.

I ran into a very annoying problem...
it seems that "some" of the linked libraries are named differently between sol8 and 10.

for example,
on i386-pc-solaris2.10
	libstdc++.so.6.0.10:
        libm.so.2 =>     /lib/libm.so.2

on i386-pc-solaris2.8
libstdc++.so.6.0.10:
        libm.so.2 =>     (file not found)
/lib/libm.so.2 does not exist on solaris 8,  it is libm.so.1


I am looking for direction,  I am now at a loss...
options at this point that I can think of are as follows:
1) use "classes" and package up both builds into a single set
	IMO not a good choice because this will double the package size.
2) package them up as separate packages.
	one set for solaris 8 and one set for solaris 10.
3) hack the code to such a point that it is all linked against the same libraries.
	I am not even sure this is possible, even if it is possible,
	I think this would result in less stable packages.


I have the test suite results from the make check in my home dir /home/mwatters/gcc-test-results/  for anyone that wants
to look at them.  they seem comparable to the ones posted at gcc.gnu.org

http://gcc.gnu.org/ml/gcc-testresults/2009-03/msg01389.html    x86 solaris 8
http://gcc.gnu.org/ml/gcc-testresults/2009-03/msg01650.html    x86 solaris10
http://gcc.gnu.org/ml/gcc-testresults/2009-01/msg02995.html  sparc solaris 8

- --
Thanks,
Mike

"Any intelligent fool can make things bigger, more complex,
and more violent.  It takes a touch of genius -- and a lot of courage --
to move in the opposite direction."

* Albert Einstein 1879 - 1955
    US German-born Theoretical Physicist
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAknGxngACgkQLrhmsXMSLxcFLACgqIqAneYXu/yqSq2/tcKn3+D/
7joAn2HoAEMRTFEU+g2fPESRv7AWUnUE
=vFRn
-----END PGP SIGNATURE-----