SUCCESS!! Many many thanx to Rainer Orth (a GCC contributor) for
nailing down the problem, offering a patch, and helping me thru a
couple of idiosycracies I have on my end.
Basically, Compaq seems to have changed their /usr/include/stdio.h
file in a manner that's unfamiliar to gcc's fixincludes program.
He offered the following patch:
----- begin patch file -------
diff -rubp gcc/fixinc/save/fixincl.x gcc/fixinc/fixincl.x
--- gcc/fixinc/save/fixincl.x Fri Sep 14 22:05:09 2001
+++ gcc/fixinc/fixincl.x Wed Nov 21 19:01:19 2001
_at__at_ -3954,6 +3954,7 _at__at_ static const char* apzStdio_Va_ListPatch
"-e", "s_at_ va_list _at_ __gnuc_va_list _at_\n\
s_at_ va_list)_at_ __gnuc_va_list)_at_\n\
s_at_ _VA_LIST_));_at_ __gnuc_va_list));_at_\n\
+s_at_ __VA_LIST__));_at_ __gnuc_va_list));_at_\n\
s_at_ va_list_at_ __not_va_list___at_\n\
s_at_\\*va_list_at_*__not_va_list___at_\n\
s_at_ __va_list)_at_ __gnuc_va_list)_at_\n\
diff -rubp gcc/fixinc/save/inclhack.def gcc/fixinc/inclhack.def
--- gcc/fixinc/save/inclhack.def Fri Sep 14 22:01:46 2001
+++ gcc/fixinc/inclhack.def Wed Nov 21 19:00:28 2001
_at__at_ -2223,7 +2223,8 _at__at_ fix = {
/*
* Use __gnuc_va_list in arg types in place of va_list.
- * On 386BSD use __gnuc_va_list instead of _VA_LIST_. We're hoping the
+ * On 386BSD use __gnuc_va_list instead of _VA_LIST_. On Tru64 UNIX V5.1A
+ * use __gnuc_va_list instead of __VA_LIST__. We're hoping the
* trailing parentheses and semicolon save all other systems from this.
* Define __not_va_list__ (something harmless and unused)
* instead of va_list.
_at__at_ -2232,6 +2233,7 _at__at_ fix = {
sed = "s_at_ va_list _at_ __gnuc_va_list _at_\n"
"s_at_ va_list)_at_ __gnuc_va_list)_at_\n"
"s_at_ _VA_LIST_));_at_ __gnuc_va_list));_at_\n"
+ "s_at_ __VA_LIST__));_at_ __gnuc_va_list));_at_\n"
"s_at_ va_list_at_ __not_va_list___at_\n"
"s_at_\\*va_list_at_*__not_va_list___at_\n"
"s_at_ __va_list)_at_ __gnuc_va_list)_at_\n"
----- end patch file -------
for the two files: fixincl.x and inclhack.def in the GCC install
source tree.
One additional caveat: If you don't have autogen install, then you
need to make sure that fixincl.x has a more recent date that inclhack.def.
So a simple "touch fixincl.x" after applying the patches will ensure
this.
Otherwise, all the other stuff that I went thru in my original attempt
seems to be unnecessary (including the changes to the as command) except
for using GNU Make. The make command installed by Compaq's Tru64 install
media is pretty darn picky. Not sure if Compaq C Compiler v6.4 had any
effect, but that thing has been nothing but trouble for me, so I may just
keep it off my machine for the time being. We do have another Tru64
machine, so when I have time, I may give it a turn over there to see if C
Compiler v6.4 has any effect.
At 5:17 PM -0800 21/11/01, I originally posted [edited for length]:
>Well, this seems to be a problem that has afflicted many others;
>unfortunately, I can't seem to dig up many summaries.
>
>After spending several hours, I finally managed to start copmiling
>gcc 3.0.2. I had to uninstall Compaq C Compiler v6.4, gcc/g++
>v2.9-gnupro-99r1 (from the OS install CDs), and manually remove
>all remaining 'stuff' from previous gcc installs. I also installed
>automake 1.5 (altho I'm not sure if that helped any).
>
>Anyway, I found one suggestions from the listserv which seemed to
>help: I created a script file /sbin/as (chmoded properly to be
>executable) which basically says:
>
>#!/bin/sh
>/usr/bin/as -oldas $*
>
>Then I configured gcc with "../gcc-3.0.2/configure --with-as=/sbin/as".
>That goes fine.
>
>Then I use GNU Make 3.79.1 (executed as "gmake") to compile the
>programs. Everything seems to go fine, until I get:
>
>gmake: *** [backspace.o] Error 1
>
>and everything errors out afterwards.
>
>[stuff deleted]
>
>The machine I'm trying to install this on is a Digital Personal
>Workstation 433au running Tru64 UNIX v5.1A (Rev. 1885) (Patch Kit 3).
--
------------------------------------------------------------------
Sung Moon Kang BK94 AA96 | UCLA - Mech & Aerospace Engineering
sung.kang.bk.94_at_aya.yale.edu | http://www.bol.ucla.edu/~kangs
Received on Tue Nov 27 2001 - 02:53:31 NZDT