update M4RI to version 20130416

[Martin Albrecht]

20130416 is mainly maintenance release of M4RI improving performance for small matrices.

Release notes are here: ​https://bitbucket.org/malb/m4ri/wiki/M4RI-20130416

Install ​http://boxen.math.washington.edu/home/malb/spkgs/libm4ri-20130416.spkg

[Martin Albrecht]

FWIW, I just built Sage 5.8 from scratch with the SPKG linked above on bsd.math.washington.edu.

[Julien Puydt]

I successfully compiled the spkg with sage 5.8 on my arm box.

[Julien Puydt]

I looked at libm4rie's compilation and found there where many warnings ; I checked libm4ri's build log, and found that many, many files trigger the following warnings:

In file included from ./m4ri/mzd.h:45:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/misc.h:577:0: warning: "__M4RI_CPU_L3_CACHE" redefined [enabled by default]
In file included from ./m4ri/mzd.h:35:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/m4ri_config.h:11:0: note: this is the location of the previous definition
In file included from ./m4ri/mzd.h:45:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/misc.h:586:0: warning: "__M4RI_CPU_L2_CACHE" redefined [enabled by default]
In file included from ./m4ri/mzd.h:35:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/m4ri_config.h:10:0: note: this is the location of the previous definitionIn file included from ./m4ri/mzd.h:45:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/misc.h:595:0: warning: "__M4RI_CPU_L1_CACHE" redefined [enabled by default]
In file included from ./m4ri/mzd.h:35:0,
                 from m4ri/brilliantrussian.h:41,
                 from m4ri/brilliantrussian.c:28:
./m4ri/m4ri_config.h:9:0: note: this is the location of the previous definition

It doesn't harm but obfuscates the logs and perhaps means there's some header file missing guards.

[Martin Albrecht]

Those warnings are expected. It means that L1-L3 were not detected successfully and are replaced by default values. So it's not critical. It might make sense to investigate how to detect L1/L2 on your kind of boxes (for future releases) and we should also print a less scary warning once.

[Volker Braun]

Looks good to me.