[GenABEL-dev] ProbABEL autotools integration branch in SVN
l.karssen at erasmusmc.nl
Mon Aug 22 16:08:18 CEST 2011
I've recently tried to create a Debian package (.deb) from the ProbABEL
source and it works! If someone on this list has experience in making
Debian packages that are fully compliant with Debian's standards we can
even try to get it included in Debian, just like plink.
I haven't tried building an .rpm yet, but that should work as well.
I propose to release a new version of ProbABEL so that other people can
take advantage of the new way the package is to be built.
I also propose to increase the version number to 0.2 to mark the fact
that a rather major change has occurred (even though users won't notice
the difference if they don't install the package themselves).
What are your thoughts?
On vr, 2011-08-12 at 15:08 +0200, L.C. Karssen wrote:
> Dear list,
> In SVN rev. 758 implemented a 'solution' for the problem with
> generating the ProbABEL PDF documentation as described in one of my
> previous e-mails (see below).
> I did the following:
> - In configure.ac I check whether the program pdflatex exists. If so a
> variable HAVE_PDFLATEX is defined.
> - In doc/Makefile.am I use this variable to add a few lines to the
> Makefile so that the PDF gets generated by 'make' and installed by 'make
> So, if you have pdflatex installed everything works like it did before.
> However, if you don't have pdflatex installed, you won't get the PDF of
> the manual, only the .tex file will be installed.
> I guess that's the best solution. At least it is self-consistent, you
> can run
> $ ./configure
> $ make distcheck
> to generate the ProbABEL-$VERSION.tar.gz file and then distribute that.
> If you untar the tar.gz file and repeat you will get the exact same
> If, on the other hand, we want our official .tar.gz file to contain the
> PDF file (which should then be generated by the maintainer of the tar.gz
> file and not stored in SVN, as Rob already pointed out), we face a
> problem, because in that case
> $ make
> $ make distcheck
> will always try to make the PDF, forcing our users to have LaTeX
> installed, which we didn't want. I tried to make the inclusion of the
> PDF into the tar.gz file optional (depending on the presence of LaTeX),
> but couldn't get that to work). I guess that's because it would violate
> the principle that it should always be possible to run make distcheck
> All in all, I think the way things are now is the most proper solution.
> Furthermore, in r. 759 I changed a bit of the ProbABEL code to reflect
> the fact that we now have the help of the configure script to give us
> information like the name of the package and the version number. The
> version number is now stored in configure.ac, which, after invocation
> of ./configure, creates src/config.h. That file contains defines like:
> #define PACKAGE_VERSION "0.1-9e"
> that I used to replace the once mentioned in version.h (which I
> deleted). A logical next step here would be to tell automake to insert
> the version number from configure.ac also into the documentation. But
> that's for another day.
> On wo, 2011-07-27 at 00:28 +0200, L.C. Karssen wrote:
> > Dear list,
> > The 'almost' I referred to earlier relates to the doc directory. At this
> > moment there is no automake Makefile.am yet for that directory. There
> > are two reasons for that:
> > 1) I'm not yet sure how to go about this. As far as I know there is no
> > automake support for LaTeX to PDF yet. However, I think I could put the
> > old Makefile rules in there with a little tweaking to make it work
> > 2) If we follow 1) and don't store the pdf in SVN, we force all users
> > who want to install ProbABEL to have a functioning LaTeX install
> > (./configure doesn't check for this AFAIK). I don't think we want to add
> > LaTeX to the ProbABEL dependencies.
> > Any suggestions on this issue is much appreciated.
> > Lennart.
> > On ma, 2011-06-27 at 19:28 +0200, L.C. Karssen wrote:
> > > Dear list,
> > >
> > > Those of you that keep an eye on the SVN repo have noticed that I've
> > > added a branch called ProbABEL/autotools-integration. In that branch I'm
> > > trying to convert the ProbABEL package to use autoconf and automake. The
> > > reasons for this change are:
> > > - most (dare I say 'advanced' here?) Linux/Unix users are used to the
> > > sequence of ./configure; make; make install when compiling and
> > > installing software.
> > > - it makes ProbABEL conform (more) to the GNU coding standards, and as a
> > > result
> > > - it is easier to package ProbABEL for various distributions
> > > (e.g. .deb, .rpm, .tgz)
> > > - I've been wanting to lean how to use autoconf and automake for some
> > > time, and ProbABEL seems to be a nice (although slightly complex)
> > > project to started with.
> > >
> > > At the moment I've got most of it working:
> > > - palinear and palogist are compiled and installed in PREFIX/bin
> > > - the examples are copied to PREFIX/usr/share/ProbABEL/examples/
> > > - documentation is copied to PREFIX/usr/share/doc/
> > > - the probabel_config.cfg file is copied to PREFIX/etc/
> > >
> > > What isn't working yet is:
> > > - the PDF is not automatically generated from the .tex file. I still use
> > > the old Makefile in the doc/ directory for that.
> > > - I didn't test how this all works for the Windows build
> > > - pacoxph doesn't compile. It gives this error:
> > > pacoxph-main.o: In function `coxph_reg::estimate(coxph_data&, int, int,
> > > double, double, int, int, int, bool, int)':
> > > /home/lennart/ErasmusMC/R-dev/ABEL-total/branches/ProbABEL/autotools-integration/ProbABEL/src/reg1.h:884: undefined reference to `coxfit2(int*, int*, int*, double*, int*, double*, double*, double*, int*, double*, double*, double*, double*, double*, int*, double*, double*, double*, double*)'
> > > collect2: ld returned 1 exit status
> > >
> > > even though the function definition seems to be available in reg1.h
> > > itself... I'll have to look into that.
> > >
> > > Any help is appreciated! Also if you've got more experience with
> > > autoconf/automake I'd like to hear your comments.
> > >
> > >
> > >
> > > Regards,
> > >
> > > Lennart.
> > >
> > >
> > > _______________________________________________
> > > genabel-devel mailing list
> > > genabel-devel at lists.r-forge.r-project.org
> > > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> > _______________________________________________
> > genabel-devel mailing list
> > genabel-devel at lists.r-forge.r-project.org
> > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
> genabel-devel mailing list
> genabel-devel at lists.r-forge.r-project.org
Department of Epidemiology
3000 CA Rotterdam
email: l.karssen at erasmusmc.nl
GPG key ID: 0E1D39E3
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 198 bytes
Desc: This is a digitally signed message part
More information about the genabel-devel