On 2007-08-18 04:13-0700 SourceForge.net wrote: > Bugs item #1776787, was opened at 2007-08-18 06:13 > Message generated for change (Tracker Item Submitted) made by Item Submitter > You can respond by visiting: > https://sourceforge.net/tracker/?func=detail&atid=102915&aid=1776787&group_id=2915 >
Arjen has already suggested on the bug tracker that this user try the cmake build system, but that still leaves the question what to do about this showstopper bug in our old autotools-based build system for 5.7.4. Here is the extent of the problem. By comparing table-of-contents for the 5.7.3 and 5.7.4 tarballs I found that cf/Makefile.in scripts/Makefile.in src/Makefile.in test/Makefile.in utils/Makefile.in are present in the 5.7.3 tarball, but missing from the 5.7.4 tarball. (I only looked for Makefile.in files, and there are probably other autotools-related files missing as well). One command that is executed by scripts/make_tarball.sh is cf/bootstrap.sh >& bootstrap.sh.out That command should generate a complete set of autotools-related files, and instead it generated an incomplete set. I ran scripts/make_tarball.sh just now on my own machine, and here is the resulting bootstrap.sh.out file: Running aclocal (GNU automake) 1.9.6... done Running autoheader (GNU Autoconf) 2.59... done Running libtoolize (GNU libtool) 1.5.22... done Running automake (GNU automake) 1.9.6...configure.ac: installing cf/install-sh' configure.ac: installing cf/missing' bindings/c++/Makefile.am: installing cf/depcomp' bindings/f95/Makefile.am: installing cf/compile' scripts/Makefile.am:37: trailing backslash on last line done Running autoconf (GNU Autoconf) 2.59... done Regenerating libltdl/aclocal+configure... done That trailing backslash error is the cause of the whole problem. I have confirmed it causes the missing Makefile.in files (and probably other missing files as well that are essential for the autotools build system to work), and that problem is corrected by getting rid of the trailing backslash (I just commited that fix to svn). My apologies for the screwup in one of my commits that failed to remove the trailing backslash when I was doing some absolutely necessary maintenance of the autotools build system. I didn't test the autotools build system afterward, because I _knew_ the change was so trivial it did not need to be tested. Live and learn. Hazen, to guard against such screwups in the future, it would be a good idea to look carefully at every *.out file produced by the scripts/make_tarball.sh script. While we are still distributing the autotools-based build system, I suppose you should do some minimal testing of it as well (such as running ./configure; make; make install by hand from the tarball version to make sure there are no obvious tarball errors such as missing files). Here is how I suggest you handle this showstopper autotools build system issue in 5.7.4. (1) Amend the 5.7.4 news item at SF and also make a note to plplot-general and plplot-devel that there is a showstopper issue in the deprecated autotools build system in 5.7.4 that we will fix in the next release, but suggest that meanwhile this is an excellent opportunity for users to build with the preferred cmake build system instead. (2) At your convenience (roughly a month from now?) bring out a release called 5.8.0-RC1 with the autotools fix that I just made as well as your matured cairo device drivers and other goodies (such as additional Ada examples) various people are working on. I think the next release should be called 5.8.0-RC1 rather than 5.7.5 as an indication it is a fairly immediate precursor to 5.8.0 and hopefully identical to it other than tag name, release numbers, etc. Note, there should be no short-cuts in this release. 5.8.0-RC1 should be a full-blown release with a unique tag name (v5_8_0_RC1), version numbers, tarball, etc., so there is no chance for confusion between 5.8.0-RC1 and its planned successor release 5.8.0. (3) Let users (and especially developers from this list) bang on 5.8.0-RC1 for a couple of weeks to shake out any obvious remaining bugs, then finally release 5.8.0. Orion, from what you have said I realize this proposed release schedule does not fit in well with the forthcoming Fedora release, but I am afraid it cannot be helped. I believe you no longer use the autotools build system so hopefully, 5.7.4 will be good enough for your rpm needs. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
