El Dissabte, 4 d'octubre de 2014, a les 23:51:04, Markus Koschany va escriure: > [dropping Barry from CC because he is subscribed] > On Fri, 03. Oct 15:43 Leopold Palomo-Avellaneda <leo@alaxarxa.net> wrote: > > Hi, > > > > > > I'm working on the ode package. I will try to propose a upload next week. > > However, after look at the package I have some doubts that I would like to > > comment: > > > > - libode has libccd embed. I did a petition to extract it to upstream and > > make a check to use the external version, if not the internal one, but no > > answer. > Hi, > > If you can use the shared system library without breaking ode, that > would be preferable. But if this is not possible for whatever reasons, > better document it in README.source with a link to the corresponding > upstream bug report. Ok, I'll do it. > > > The build system is autotools and I'm not so expert to make a patch. So, > > can we provide libode with this code embed? > > We have been doing this for the past five years, so probably the answer is > yes but...(see the paragraph above) ok > > - Upstream has changed SONAME. Also, we have had two packages, one with > > single precision test and another with double precision. I think that we > > could drop the single precision package and release the double precision > > as default in 64bit arch and single precision in 32bits arch. What do you > > think? > A SONAME change means changing the name of the binary package too. That > would require a transition for all reverse dependencies but the > timeframe for transitions is already closed for Jessie. Of course you > could upload a new version to experimental. About single- vs. double > precision: Your suggestion sounds reasonable but are there any r-deps > that still require the single-precision package? Why was this package > created in the first place? Well, I would like to upload to Jessie this version. I didn't known the question of transitions closed, but libode has this r-dependencies: The easiest: $ apt-cache rdepends libode1sp libode1sp Reverse Depends: libode1sp:i386 python-soya-dbg python-soya libode-sp-dev mokomaze --> mokomaze and python-soya needs the single precision version. I can ask to the maintainers. the difficult ... leo@indiana:~$ apt-cache rdepends libode1 libode1 Reverse Depends: k3d libode1:i386 xmoto libtaoframework-ode0.9-cil stormbaancoureur python-pyode python-pyepl libompl9 libode-dev mu-cade k3d darkplaces-server darkplaces libcrystalspace-2.0 if I ask to the maintainers and comment the situation and if they agree it could be possible to upload the new packages? > > > - I think that we can drop dh-autoreconf. It was introduced in the > > previous > > version, not released in debian. I would prefer it, because not, recreate > > the autotools files and it's a mess when the devscripts make the diff > > with the orig version. What do you think? > > dh-autoreconf is usually the preferred method when building packages > with autotools. I suggest to convert debian/rules to dh-sequencer and to > use the --with autoreconf option. I don't understand the last sentence. > Shouldn't dh_autoreconf_clean remove all files that are recreated during > the autoreconf step? No, maybe I have not been clear and probably it has been my fault. This is the problem: - Upstream release a sources with config.guess, configure, aclocal.m4, aclocal.m4, Makefile.in, etc files. T - I work on the package in a chroot with sid and I do a debuild. As rules regenerate this file, the original sources are different so it complains that I have a different version of sources. And the only difference is that the original ones are for instance: -[AM_AUTOMAKE_VERSION([1.14])dnl +[AM_AUTOMAKE_VERSION([1.14.1])dnl dh $@ --with autoreconf is who do that. If you explain me another way to solve it, perfect. > > - Someone knows why 0.12 was never uploaded to unstable? > > Sorry, no idea. Probably the usual lack of time to complete the upload. > Ok, Thanks, Leopold -- -- Linux User 152692 PGP: 05F4A7A949A2D9AA Catalonia ------------------------------------- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?
Attachment:
signature.asc
Description: This is a digitally signed message part.