Hi,
The answer would appear to be to pay a little more attention to the build
instructions. Answers were sent by
Paul A Sand [pas_at_unh.edu]
Greg Freemyer [freemyer_at_NorcrossGroup.com]
John Losey [JLosey_at_Solucient.com]
Dave Love [d.love_at_dl.ac.uk]
redhatcat [redhatcat_at_fastdial.net]
Of those the one that hit the spot was from Paul who condensed the install
instructions to
gunzip < gcc-3.0.4.tar.gz | tar xf -
mkdir obj
cd obj
env CC=cc ../gcc-3.0.4/configure
gmake bootstrap-lean
gmake install
I had created an object directory, but I hadn't made sure I was using gmake
until I renamed /usr/local/bin/make to gmake.
I have since compiled a couple of small g++ programs and am now in the
process of compiling apache/mysql/php with oracle as a reasonable road test.
Dennis
The original question:
> I have been trying to compile gcc3.0.4 for the last day or so. What a
chore
> this is turning out to be. 2.95 was fairly monumental as well. This time
the
> make is having trouble sorting out stdc++. I'm on a DS10 ev67 with Tru64
> v5.1A.
>
> I have since gone back and looked at my compile script and looked at the
> installation instructions and tried the whole thing again. As yet I
haven't
> come up against the stdc++ problem. However, I feel I am going to at some
> stage. I welcome anyone who has some gen on how I can generate a
> libstdc++.so or libstdc++.a. After my initial attemps to build gcc/g++,
and
> attempts to force the build libstdc++ from the gcc source directory which
> included a libstdc++-v3 directory, turned to copying down
> libstdc++-v3.0.97-gcc-3.0.4.tar.gz, thinking I could use that. However
that
> also failed for the following reason
> creating cache ./config.cache
> configure: error: can not find install-sh or install.sh in
> /usr/local/src/libstdc++ /usr/local/src/libstdc++/..
> /usr/local/src/libstdc++/../..
>
> Here are some or all of the steps of my latest attempts to compile gcc.
>
> I renamed anything in /usr/local/bin that I thought was related to gcc,
g++,
> g77, gj to {}.bup. I was thinking I needed to force the use of cc in the
> bootstrap.
>
> The following gz files were unpacked in the same directory
> gcc-3.0.4-20020215.tar.gz
> gcc-core-3.0.4-20020215.tar.gz
> gcc-g++-3.0.4-20020215.tar.gz
> gcc-g77-3.0.4-20020215.tar.gz
> gcc-java-3.0.4-20020215.tar.gz
> gcc-objc-3.0.4-20020215.tar.gz
> gcc-testsuite-3.0.4-20020215.tar.gz
>
> Previously I had to apply a couple of patches at this point,
> ?? Apply patch - duplicate symbols to the gcc (Y/N) [N]:
> ?? Apply patch - unresolved ASM_FILE_START to the gcc (Y/N) [N]:
> I disregarded these patches, this time around.
>
> Then I cd'ed into the unpacking directory and issued "./configure
> --prefix=/usr/local --enable-shared". Not quite sure why I wanted
> --enable-shared, it was in the script from the compile of 2.95, so I just
> left it in. That seemed to work OK.
>
> Then "make boostrap". That seemed to be working fine until it hit
> machmode.def at which time it put out the warning
> cc: Warning: /usr/local/src/gcc/gcc/machmode.def, line 73: In the
> initializer for mode_mask_array[5], the shift count
> "64" is negative or is greater than or equal to the promoted size of
> the operand "(unsigned long)1". (shiftcount)
> DEF_MACHMODE (DImode, "DI", MODE_INT, BITS_PER_UNIT*8, 8, 8, TImode)
> I have no idea what this ones about.
>
> After a few messages the compile gave up, with a network disconnect
message.
> I guess that is not related to the compile problem, but I'm wondering if
the
> above is going to cause problems, it sort of feels like the compiler
should
> know what it is doing at the bit/byte level.
######################################
Dennis Macdonell
Systems Administrator
National Mapping Division, Geoscience Australia
mail: PO Box 2, Belconnen, ACT 2617
email: mcdonell_at_auslig.gov.au
ph: 61 2 6201 4326
fax: 61 2 6201 4377
######################################
Received on Thu Mar 28 2002 - 00:07:08 NZST