From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 27957 invoked by alias); 7 Feb 2003 03:22:46 -0000 Mailing-List: contact insight-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: insight-owner@sources.redhat.com Received: (qmail 27950 invoked from network); 7 Feb 2003 03:22:46 -0000 Received: from unknown (HELO redhat.com) (66.30.22.225) by 172.16.49.205 with SMTP; 7 Feb 2003 03:22:46 -0000 Received: by redhat.com (Postfix, from userid 201) id 1A4A61B8E7; Thu, 6 Feb 2003 22:23:37 -0500 (EST) Date: Fri, 07 Feb 2003 03:22:00 -0000 From: Christopher Faylor To: insight@sources.redhat.com Subject: [RFC] Remove tcl/tk/itcl/itk dependencies from Makefile.in Message-ID: <20030207032337.GA29239@redhat.com> Mail-Followup-To: insight@sources.redhat.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.1i X-SW-Source: 2003-q1/txt/msg00087.txt.bz2 In the neverending process of dealing with my favorite cygwin activity, I decided that it would be best if I completely removed any tcl/tk stuff from the source directory of the insight that I build. Instead, I wanted to use the --with-tclconfig and --with-tclinclude options to point to another directory. In the process of doing this, I found that there is a need for a --with-itclconfig and (maybe) a --with-itkconfig. Both options are missing from configure.in. Lack of those is easy enough to workaround by priming config.cache but I'm wondering if it would be worthwhile to implement them? A larger problem however is the fact that currently, the options don't completely work, since Makefile.in has settings for TCL_DEPS and TK_DEPS which assumes that they live in gdb's build tree, e.g., in ../tcl/win/libtcl84.a. This obviously isn't the case when using a separate tcl/tk build. The fix for this is really simple, however. Remove the TCL_DEPS style settings and just let the gdb link happen. This means that you get a linker error rather than a Makefile error but is that really a big deal? If gdb automatically rebuilt the libtcl.a file when it was out-of-date, I could see a reason for the dependency. It doesn't do that, though. So, presently you get the potentially more informative error message: *** No rule to make target `../tcl/win/libtcl84.a', needed by `gdb.exe'. Stop. rather than (I propose): ld: cannot find -ltk84 but the make error is wrong in the case where --with-tclconfig is used anyway. And, if you're building gdb from the top-level, it's unlikely that libtcl*.a will not be built. Anyway, I've managed to build a version of gdb by removing TCL_DEPS, TK_DEPS, ITCL_DEPS, and ITK_DEPS from the Makefile and I can kludge something for my cygwin build system so that I can make this automatic. I was just wondering if anyone would mind eliminating these dependencies. If the tclConfig.sh file provided anything useful which could be extrapolated to put into the TCL_DEP, I'd use that but it doesn't seem to do that. It only provides a path to the library using -L/wherever -ltcl84. I could munge that together into /whatever/libtcl84.a but that seems sort of fragile. So, the question is: Is it ok to remove the tcl/tk/itcl/itk *_DEPS files from the gdb Makefile? This would also be a step in the direction of allowing insight to be built from system tcl/tk libraries. cgf