lp:ubuntu/trusty-proposed/atlas

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

28. By Sébastien Villemot on 2014-02-03

* libatlas3gf-base: update long description to suppress lintian warning.
* Remove architectural defaults for obsolete s390 arch.
* Use the new GENERIC arch name for powerpc, instead of POWER3.
  Hopefully that will fix the FTBFS on powerpc.

27. By Sébastien Villemot on 2014-01-29

* libatlas-base-dev now provides liblapack.so.
* Remove unused libatlas-3.so, libatlas-3gf.so and libatlas.so.* virtual
  packages.
* Rewrite logic in libatlas3-base.postinst. (Closes: #719142)
* Disable autodetection of machine type on archs for which ATLAS does not
  provide generic-enough architectural defaults. This is achieved by creating
  a GENERIC machine type (see generic.diff). In particular, this ensures that
  the package created on an armel buildd with an ARMv7 will be generic.
  (Closes: #719355)
* Register documentation to doc-base.
* Bump Standards-Version to 3.9.5, no changes needed.

26. By Sébastien Villemot on 2013-07-27

* Allow the generic package to build on machines with CPU throttling
  enabled. Otherwise the package FTBFS on some buildds (e.g. biber).
  Implementation is done by reactivating the "-Si cputhrchk 0" flag
  (cpu-throtthling-check.diff), and using it in debian/rules.
* Add architectural defaults for armel and mips.
* armhf.diff: do not enforce 32-registers FPU for Fortran

25. By Andreas Beckmann on 2013-04-28

* Non-maintainer upload.
* libatlas3gf-base: Add Breaks: octave3.2 (Closes: #706328)
  This fixes some upgrade paths by removing the obsolete octave3.2 (and its
  triggers) a bit earlier. This is a workaround for dpkg bug #671711: dpkg
  may run trigger processing for a package even if its dependencies are not
  satisfied. The octave3.2 triggers may be run in such a state (with
  liblapack* and libblas* unpacked but unconfigured, therefore the
  alternatives may not be setup or dangling) and will fail, aborting the
  upgrade, even if octave3.2 would be removed later anyway.
  Similar Breaks exist in libblas3 and libarpack2 to cover more incarnations
  of this bug.

24. By Sébastien Villemot on 2012-08-04

[ Sébastien Villemot ]
* libatlas-test: put in devel/extra, to reflect override
* Force generic PPC CPU when building a non-custom package on a PPC arch
  (Closes: #680096)

[ Sylvestre Ledru ]
* Use series for list the patches
* Patches refreshed

23. By Sébastien Villemot on 2012-06-24

* libatlas3-base.prerm, libatlas3gf-base.prerm: avoid crash if libblas.so.3gf
  or liblapack.so.3gf already deleted
* Add 3gf compat symlinks for libcblas, libf77blas and liblapack_atlas
  (to avoid breakage of some reverse deps)
* In the LAPACK alternatives, use a lower priority than Netlib (in order to
  minimize the incidence of #576972)
* Update and clean-up README.Debian
* Mention known build failures in README.source
* libatlas3gf-base transitional package:
  + move to section oldlibs
  + mark as arch:all
* Add myself to Uploaders
* Set DM-Upload-Allowed to yes

22. By Sébastien Villemot on 2012-06-13

* Force generic x86 CPU when building a non-custom package on an x86 arch
  (Closes: #676885)
* Force 2 threads when not building a custom package (as was the case up
  to and including 3.8.3-30)
* Fix detection of some Intel Core i1 CPUs (corei1.diff)

21. By Sylvestre Ledru on 2012-06-08

Yet an other stupid bug...
Merci à Rafael Laboissiere (Closes: #676614)

20. By Sylvestre Ledru on 2012-06-07

Bring back slave alternative for liblapack.so.3gf.
Merci à Sébastien Villemot (Closes: #676495)

19. By Sylvestre Ledru on 2012-06-02

Upload in unstable (Closes: #673357)