gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] gcl compilation failure


From: Camm Maguire
Subject: Re: [Gcl-devel] gcl compilation failure
Date: 16 Dec 2001 11:33:28 -0500

Hi Vadim!

"Vadim V. Zhytnikov" <address@hidden> writes:

> This is a multi-part message in MIME format.
> --------------DEA2387FD7C934F3F50BC005
> Content-Type: text/plain; charset=koi8-r
> Content-Transfer-Encoding: 7bit
> 
> I've just tried to build gcl first dynamically and later
> statically
> on my Linux box (ALT Linux Sisyphus, Pentium III, Kernel 2.4.12,
> glibc 2.2.4, gcc 2.95.3). Unfortunately in both cases make fails
> with virtually the same diagnostics (logs are attached). Any
> advises
> on how to track real source of the problem down are greatly
> appreciated.
> I'm ready to make any test to clarify situation but at present I
> do not
> know what to do. Strace, ltrace?
> 

OK here are your problems, I think:

checking for wish8.0... (cached) wish
Application initialization failed: no display name and no $DISPLAY environment 
variable

This results in your TCL/TK variables being empty:
TK_CONFIG_PREFIX=
TK_LIBRARY=
TCL_LIBRARY=
TK_XINCLUDES=
TK_INCLUDE=
TCL_INCLUDE=-I/usr/lib/../include


Later, you show your failures only at the end when you try to load
your tkl.o module with the Tk GUI interfase.  Your gcl compiles other
stuff successfully several times before.  So something is wrong with
your TK setup.  Please try:

1) Running configure under X with a correctly set DISPLAY variable
2) Try using instead the configure flags:
        --enable-tkconfig=<dir of your tkConfig.sh>
        --enable-tclconfig=<dir of your tclConfig.sh>

We definitely don't want the build to only work under X, and need to
rework this, but lets see if we can pinpoint the failure first.

> A also send gcl-readline.patch. I think it is worth including in
> cvs.
> It works to me.
> 

Thanks!  This is a definite add in my book too.  First, though, I'd
like to get a working 2.4.1 released with no new features, only bug
fixes/build fixes.  Then we can follow the linux kernel pattern and
open up 2.5.0 as "unstable" :-)


> Finally I've noticed a little bug in the configure script.
> When "buggy gcc 2.96" is checked the script tests gcc
> directly. It should check $GCC (compiler passed to
> configure as the parameter) instead. The point is that
> on RH, Mandrake and derivatives gcc is usually 2.96
> but gcc 2.95.3 is invoked as kgcc.
> 

OK, I've just committed a configure.in/configure rework of the CC
handling.  Looks like GCC was just substituted in the case of
386-linux, but it definitely appears to be needed in the makefiles in
all cases.  You should be able to select the compiler now through the
CC environment variable, i.e. CC=??? ./configure -- to my understanding
you could not do this before -- it was hardcoded to gcc.

On my todo list, though I don't know where, is to 

1) rework the configure.in, making use of the available autoconf
   macros in places where custom scripts have been written
2) Get rid of the sed/makedefs replace scheme, and get proper
   Makefile.in files in all subdirs.
3) Get good clean and distclean targets to the make.
4) Have compiler selected in the normal fashion, via --with-gcc= to
   configure.  Treat other key variables similarly.

> Vadim
> 
> --
> 
> [ Vadim V. Zhytnikov  <address@hidden>  <address@hidden> ]
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah



reply via email to

[Prev in Thread] Current Thread [Next in Thread]