The problem, of course, is that it requires == 1.6 on both of them. I have 1.7 installed. I had to symlink $HOME/bin/automake-1.6 to /usr/bin/automake-1.7, and same with aclocal, to get it to work. (This is from CVS on sourceforge.)
I ended up having to abort the install because I don't have yacc or bison installed (I'm using a shared box that I'm not the admin of, and have a rather restrictive quota on). I'm getting one (or both) of them installed in the near future; once that happens I'll probably be able to build it. But I can't for the life of me figure out why it required == 1.6. -Kyle -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of David Evans Sent: Tuesday, January 13, 2004 8:04 AM To: [EMAIL PROTECTED] Subject: Re: [splint-discuss] Why require specific versions of automake and aclocal? (fwd) I asked Tim Van Holder, who contributed the autoconf and automake build files for splint, to answer these questions. His explanation is below. If anyone can contribute further improvements to the build process, they are appreciated. > ---------- Forwarded message ---------- > Date: Tue, 13 Jan 2004 10:46:26 +1300 > From: Simon Hosie <[EMAIL PROTECTED]> > Reply-To: [EMAIL PROTECTED] > To: [EMAIL PROTECTED] > Subject: RE: [splint-discuss] Why require specific versions of > automake and > aclocal? > > Kyle Hamilton: > > Why does splint require specific versions of automake and aclocal > > (specifically, 1.6 for each)? > > I'd go a step further and ask why splint requires auto* at all. > > Why does splint require auto* at all? Hmm - I'm not sure - I don't remember requiring 1.6 specifically. In any case, if there is such a test, then it probably requires 1.6 *or higher*. Given that I was using 1.6 at the time it seems plausible I added the 'at least 1.6 needed' requirement to avoid issues if a feature I used did not exist in earlier versions. Also, automake is up to 1.7.x now (possibly even 1.8), so I don't think there's a big problem here. Of course, if the requirement is lowered to 1.5 or 1.4 and everything still works, then that's fine. Actually, scratch all that - I just checked the CVS repository at splint.sf.net. Neither configure, nor Makefile.am, requires automake 1.6. configure simply requires automake (any version) and the Makefile.am files specify that automake 1.5 or higher is required (and everyone should have automake 1.5 or higher). However, the CVS repository also contains automake output, and automake 1.6 and up have the understandable but annoying habit to hardcode their own version into the makefiles (so that you can have automake 1.6, 1.7 and 1.8 installed on the same machine without version conflicts with its runtime, and the package will keep using the same automake version until you rerun configure). Whether or not automake and autoconf output should be included in CVS is a (mostly) a matter of taste - my opinion is that if someone gets a product out of CVS they can be reasonably expected to have some minimal development environment set up. In fact, that's what the bootstrap script is for - after a CVS checkout, you run it to have autoconf & co run and prepare the tree for compilation. In source tarballs, you can safely include the autoconf & automake output - a non-developer will never modify configure.ac or Makefile.am and so is never faced with having to run any version of autoconf or automake. So I would suggest removing configure, aclocal.m4 and all automake generated Makefile.in files from the CVS repository. Alternatively, add a note to README that running the bootstrap script is recommended after a CVS checkout. In case this message came from someone building from a source tarball without having modified any of the autoconf or automake input, there may be some flaw in the splint release procedure - but a make dist (which I assume is used) normally ensures everything builds, so that seems unlikely. If there are any further questions, feel free to ask. _______________________________________________ splint-discuss mailing list [EMAIL PROTECTED] http://www.splint.org/mailman/listinfo/splint-discuss _______________________________________________ splint-discuss mailing list [EMAIL PROTECTED] http://www.splint.org/mailman/listinfo/splint-discuss
