From: Rafael L. <rla...@us...> - 2003-03-05 14:59:20
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv10678 Modified Files: sysloc.in Log Message: After Joao's suggestion, improved the way the plplot_octave-libdir.oct file is generated, such that it is not doubly compiled (should also work for octave 2.0). The check for flag --compile of mkoctfile is removed from sysloc.in, as well as the AM conditional mkoctfile_accepts_c_flag. Also, plplot_octave_txt/* is added to the list of CLEANFILES. |
From: Rafael L. <rla...@us...> - 2003-03-05 15:02:43
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv13190 Modified Files: Makefile.am Log Message: Added sysloc.in to EXTRA_DIST. |
From: Rafael L. <rla...@us...> - 2003-03-06 00:42:49
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv10951 Modified Files: configure.ac Log Message: Switched : plplot.pc.in -> plplot-float.pc.in plplot-double.pc.in -> plplot.pc.in This reflects the fact that we have now with_double=yes in configure.ac. |
From: Rafael L. <rla...@us...> - 2003-03-06 15:34:17
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv26299 Modified Files: configure.ac Makefile.am Log Message: This is the brute force approach to get "make dist" to include (almost) all the files that are present in a freshly checked out CVS tree. The notable exceptions are the directories cf/, new/, tmp/, and bindings/perl5. This was essentially done by putting files in the EXTRA_DIST variable in the Makefile.am files. I really dislike this brute force method, but since we really need to make a release soon, I see no way of doing otherwise. I would prefer to take this opportunity to remove some legacy cruft and to improve our installation in a file-per-file basis. For instance, in directory bindings/tcl, I added blindly the files README.tclAPI and plitclgen to EXTRA_DIST, but it is clear that they could be installed somewhere with "make install". Same commentary for the Python and Java stuff. Since I am (almost) totally ignorant on Python, Java and Tcl/Tk, I will really appreciate if competent developers step in. This cleanup + installation improvement discussion must take place in the post-release times. An aside note: the EXTRA_DIST variable (or the variables it uses) should not be inside conditionals in Makefile.am, otherwise the distribution tarball may be wrong. This is not yet consistently done throughout the source tree. |
From: Rafael L. <rla...@us...> - 2003-03-12 19:54:30
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv14287 Modified Files: configure.ac bootstrap.sh Log Message: Added options --version=VER, --date-version, and --help to bootstrap.sh. Added a comment in configure.ac to remember the developers to not mess up with the format of the AM_INIT_AUTOMAKE line. |
From: Rafael L. <rla...@us...> - 2003-03-12 22:32:27
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv26906 Modified Files: configure.ac Log Message: The AM_CONDITIONALs inside the macro CHECK_PROG (in doc/docbook/configure.ac) were not being defined when --disable-docbook was given to configure. The only relevant AM_CONDITIONAL (namely has_onsgmls) has been moved outisde the test for enable_docbook in configure.ac. |
From: Alan W. I. <ai...@us...> - 2003-03-13 06:22:48
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv31197 Modified Files: configure.ac Log Message: Change version number to 5.2.0.cvs as discussed on list. |
From: Rafael L. <rla...@us...> - 2003-03-13 19:54:54
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv24857 Modified Files: configure.ac Log Message: Implemented the check of programs and Perl modules necessary to build the files in bindings/octave/plplot_octave_txt/. The extension of the present change is quite large, because the previous situations was quite constraining. Indeed, the checks for Perl modules were buried in bindings/octave/configure.ac, which was included in configure.ac conditionally on enable_docbook. To make the idea work, I had to split bindings/octave/configure.ac into two parts: the first one consisted of macro definitons, configure options declarations and things alike, that can (and sometimes must) run inconditionally. This material has been moved into a new file called bindings/octave/docbook.m4 (BTW, having two configure.ac in the source tree was an aberration). The second part consisted of the actual checks (for programs, DTDs, and Perl modules), which should be run conditioanlly. This material has been inserted into configure.ac. The specific tests for the Octave bindings regard the program Perl, and the Perl modules XML::DOM and XML::Parser, which are used by api2text.pl to build the text documentation files. If the tests go okay, api2text.pl is called in bindings/octave/Makefile.am, if needed. Otherwise, thanks to the new AM_CONDITIONAL has_xml_pm, if recompilation of the *.txt files is needed, then the make and shows the error message: Perl modules not available: cannot generate plplot_octave_txt/*.txt This has been tested in trivial cases, but must still be made bullet-proof. |
From: Rafael L. <rla...@us...> - 2003-03-14 00:20:23
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv6874 Modified Files: sysloc.in Log Message: Added AC_SUBSTitution of variable OCTAVE_VERSION. This has no impact in the current version of the configuration scheme, but I am adding it for completeness sake. |
From: Rafael L. <rla...@us...> - 2003-03-14 01:49:54
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv3855 Modified Files: sysloc.in Log Message: Made OCTAVE_VERSION store just the version number, and not the (<arch>) that was appended before. |
From: Rafael L. <rla...@us...> - 2003-03-14 09:44:13
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv6116 Modified Files: sysloc.in Log Message: The logic for checking the swig verison number is correct now. If the swig program was unavailable, configure would die with: ./configure: line 12163: < 17: syntax error: operand expected (error token is "< 17") This fixes Debian Bug#184694. |
From: Rafael L. <rla...@us...> - 2003-03-14 11:53:05
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv26280 Modified Files: configure.ac acinclude.m4 Log Message: Excerpt of acinclude.m4, regarding this change: dnl ------------------------------------------------------------------------ dnl Make AC_OUTPUT smarter dnl CFFILES_SAVE dnl AC_OUTPUT dnl CFFILES_RESTORE_MAYBE dnl The macros below prevent needless touching of <file> when AC_OUTPUT dnl makes AC_SUBSTitution of <file>.in. This will avoid unnecessary dnl recompilations when <file> is in the list of dependencies in a Makefile dnl rule. These macros should be put around the call to AC_OUTPUT. They rely dnl on the autoconf variable ac_config_files, set by the AC_CONFIG_FILES dnl macro. dnl dnl Since this is an experimental feature, a new configure option is dnl introduced (--enable-smart-ac-output) which defaults to "no". Hence, dnl without specifying this option, the behavior of configure is the same dnl as before. Alan, if this works as expected (and it does seem to work), then you can go and change all my "sloppy style" dependencies in the Makefile.am files! |
From: Alan W. I. <ai...@us...> - 2003-03-14 16:45:41
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv1125 Modified Files: bootstrap.sh Log Message: Fixed inconsistent spelling bug: aclocal_options ==> aclocal_opts |
From: Rafael L. <rla...@us...> - 2003-03-14 22:53:53
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv10266 Modified Files: sysloc.in Log Message: This is the implementation of Joao's idea: if the Octave prefix does not coincide with the PLplot one, then the plplot_octave.oct file is installed in the same directory of the .m scripts (PLPLOT_OCTAVE_DIR). Otherwise, it gets installed in the Octave versioned directory present already in the system (${prefix}/libexec/octave/<version>/oct/<arch>). This has the nice effect of really limiting the use of PLplot_Octave to the version of Octave with which the bindings have been built. I recently knew that there are binary incompatibilities between the .oct files generated with mkoctfile-2.1.36 and mkoctfile-2.1.46. |
From: Rafael L. <rla...@us...> - 2003-03-16 22:44:34
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv25095 Modified Files: Makefile.am Log Message: Added config.status to EXTRA_DIST list. This is completely harmless in normal operation (since this file is regenerated at each call to configure), but its inclusion in the tarball is essential to avoid unneeded recompilation of the docs. |
From: Rafael L. <rla...@us...> - 2003-03-17 15:32:28
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv1351 Modified Files: configure.ac acinclude.m4 Makefile.am Log Message: I think that I finally got right the avoidance of compilation of the DocBook files after running configure. There is now a new configure option, namely --disable-docbook-build, which is set to "no" by default (the normal behavior is thus preserved). There is a new AM_CONDITIONAL (docbook_build, defined in doc/docbook/docbook.m4) that controls building in doc/docbook/src/Makefile.am. All my previous hacks involving AC_OUTPUT and config.status are not needed anymore anmd have been removed from configure.ac and acinclude.m4. There is no need for including config.status in the tarball, and its line has been removed from EXTRA_DIST in Makefile.am. |
From: Alan W. I. <ai...@us...> - 2003-03-17 16:08:31
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv19344 Modified Files: configure.ac Log Message: Remove ability of user to modify PYTHON_INSTDIR. This symbol is only used in examples/python/plplot_python_start.py.in, but any deviation from the $pythondir value (with actual prefix replacing the symbolic prefix used in that symbol) would break examples/python/plplot_python_start.py. |
From: Rafael L. <rla...@us...> - 2003-03-19 21:44:34
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv11834 Modified Files: configure.ac Log Message: Implementation of Alan's idea for better control of Docbook documentation check/build. The previous options --enable-docbook and --enable-docbook-build have been replaced by: --enable-builddoc Enable configuration checks and build of DocBook documentation. Default value is "no". --with-prebuiltdoc Assume that the DocBook documentation is already built and present in doc/docbook/src, ready to be installed. This option is useful for package maintainers. Default value is "no". Specifying both options at the same time is logically inconsistent and will result in an error message with configure been stopped. The implementation was quite straightforward, because I just wrapped the two AM_CONDITIONALs enable_docbook and docbook_build around the two configure options above. A nice improvement over the previous situation is that the Docbook checks are not exercised when the docs are already present and needs only installation (i.e. when using --with-prebuiltdoc). |
From: Alan W. I. <ai...@us...> - 2003-03-21 19:13:31
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv2094 Modified Files: configure.ac Log Message: Define new symbol PLPLOT_TCL_VERSION=5 and use it in the various configured pkgIndex.tcl files. |
From: Rafael L. <rla...@us...> - 2003-03-23 23:57:04
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv31672 Modified Files: acinclude.m4 configure.ac Log Message: Added the definition of macro CHECK_GNU_MAKE in acinclude.m4 and its call in configure.ac. This macro checks for the GNU make program. For now, no action is taken, but this may be used in the future for disabling the documentation building, since the make-GNUisms are not going to be removed from doc/docbook/src/Makefile.am. |
From: Alan W. I. <ai...@us...> - 2003-03-24 05:03:35
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv23378 Modified Files: configure.ac Log Message: Fix ordering bug (perl must be checked before perl modules) that was introduced recently with --enable-builddoc changes. |
From: Alan W. I. <ai...@us...> - 2003-03-24 06:29:52
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv12448 Modified Files: configure.ac Log Message: Refine the perl, XML::DOM and XML::Parser check logic so it will not occur when the documentation is prebuilt. |
From: Rafael L. <rla...@us...> - 2003-03-24 09:14:18
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv30958 Modified Files: configure.ac Log Message: Made check for perl always occur in configure (at Alan's request). I agree that this seem necessary, since perl is called to build the tclAPI stuff in the drivers/ directory. I have never noticed that and found very strange that we have this duplication of code in respect to bindings/tcl. I am almost sure that it must be an elegant way to avoid this duplication. Also, for consistency, changed "perl" to "$(PERL)" in drivers/Makefile.am. |
From: Alan W. I. <ai...@us...> - 2003-03-24 22:29:14
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv26275 Modified Files: configure.ac Log Message: Append $X_CFLAGS to TKINCCMD. (X_CFLAGS is defined on all platforms, and some platforms [netbsd for one] need the X include location part of the TKINCCMD string in order to compile anything that includes the Tk header). |
From: Rafael L. <rla...@us...> - 2003-03-25 11:28:06
|
Update of /cvsroot/plplot/plplot In directory sc8-pr-cvs1:/tmp/cvs-serv28401 Modified Files: configure.ac Log Message: This set of changes should fix some (if not all) of the remaining problems with the tarball generation related to built files generated with Perl. The $PERL variable is now set the a warning message in configure.ac, such that if Perl in not found by configure, make will not die furiously. I hope I traced down all the places where Perl scripts are invoked in the Makefile.am files and I prepended $(PERL) to the calls. There may be some others remaining places still (excepted doc/docbook), though. With the current changes, regular users building from the tarball do not need Perl available in their systems in order to build and install PLplot. The last problem that I noticed was with the generation of the bindings/octave/plplot_octave_txt/*.txt files. The doc-stamp target has been brought outside the AM_CONDITIONAL enable_octave and the EXTRA_DIST variable is not set conditionally anymore. Also, fixed the macro CHECK_PORG in doc/docbook/docbook.m4, which is used to detect the Perl program. The variable PROG is now correctly set to the null string if the program is not detected. Unrelated to the above, a check for the libm is introduced before the check of functions isinf, finite, and isnan. This should fix the configure problem in Alpha-OSF. |