Permalink
Commits on May 27, 2010
  1. Bump version

    committed May 27, 2010
  2. Add some typemaps for CBLAS

    committed May 27, 2010
  3. Fiddle with error handlers and such

    Attempt to fix the test suite to deal with most of these issues:
    
    http://www.cpantesters.org/cpan/report/7328313
    committed May 27, 2010
  4. More is -> is_similar in t/Linalg.t

    committed May 27, 2010
  5. is -> is_similar

    committed May 27, 2010
Commits on May 26, 2010
  1. Improve typemap and malloc slightly more memory, which makes the Sort…

    … tests work, illvilja++
    committed May 26, 2010
  2. Improve .gitignore

    committed May 26, 2010
Commits on May 18, 2010
  1. Bump minor version number

    committed May 4, 2010
Commits on May 17, 2010
  1. Comment out some more free() calls, which need to be turned on per-su…

    …bsystem to weed out memory bugs
    committed May 17, 2010
Commits on Apr 20, 2010
  1. Trade memory leaks for segfaults

    committed Apr 20, 2010
Commits on Mar 10, 2010
Commits on Mar 9, 2010
  1. Give Thierry proper credit

    committed Mar 9, 2010
  2. Merge branch 'master' into bleed

    Conflicts:
    	bin/smolder_mathgsl
    committed Mar 9, 2010
Commits on Dec 1, 2009
  1. Added myself to the CREDITS file (after Jonathan Leto reminded me abo…

    …ut it)
    Jakob Ilves committed Dec 1, 2009
  2. In instantiation of GSLBuilder in Build.PL, the value for the paramet…

    …er include_dirs were corrected to an anonymous empty array, instead of an empty string.
    
    There were a build failure where gcc gets confused and 'loses' one
    of the supplied -I flags (the one specifying the directory
    containing the core perl header files), resulting in gcc not
    finding a number of very fundamental header files for perl modules.
    
    This is caused by a single '-I' flag appearing before the 'lost' -I
    flag.  That lone '-I' flag is the result of trying to use the -I
    flag to add a directory consisting of the empty string to the list
    of directories to search for include files.
    
    The fix is to prevent that bogus empty string directory to enter
    the build logic in the first place:
    
    In the instantiation of the GSLBuilder class, I changed the
    value passed for the 'include_dirs' from an empty string to a
    (reference to) an empty array.  This prevents the lone '-I' flag
    to be passed to gcc, making gcc properly understand the rest
    of the command line.
    
    (The GSLBuilder code can be amended to actually fail when
    'include_dirs' is given an empty string, and in that failure,
    tell that 'include_dirs' should be given an empty list if
    one wants to state that there should be no include directories)
    Jakob Ilves committed Nov 30, 2009
Commits on Nov 23, 2009
  1. @pjotrp

    Typo

    pjotrp committed with May 23, 2009
  2. @Moisan
  3. @Moisan

    Add == and != operators to Vector

    Moisan committed with Nov 18, 2009
  4. Fix hypergeometric functions not included in SF

    Thierry Moisan committed with Nov 12, 2009
Commits on May 30, 2009
Commits on May 27, 2009
  1. Revert 12da081

    committed May 27, 2009
  2. make smoke_some_mathgsl find a perl

    committed May 27, 2009