Le mer 14/07/2004 à 10:29, Aleksey Gurtovoy a écrit :

Michael Stevens writes:

...

...

uBLAS Done

Michael, does it mean that the CVS contains the updated uBLAS code? Note that according to http://tinyurl.com/3vwcz there is a significant number of regressions on Borland and CodeWarrior. I'd appreciate if somebody could look into these.

Speaking of uBLAS, test7 will never be able to pass a regression test, whatever the compiler/platform. Indeed, in order for this test to even compile, the macro BOOST_UBLAS_USE_INTERVAL needs to be defined. The Jamfile defines such a macro in the "exe test7" rule, but this rule doesn't seem to be taken into account when regression testing. So the Jamfile of uBLAS should be fixed, or the testing Jamfiles should be enhanced, or... Please note that it is the same situation for all uBLAS Jamfiles (even if test7 is the only one to fail because of it, I think). Regards, Guillaume

Vladimir Prus wrote:

Guillaume Melquiond wrote:

...

Speaking of uBLAS, test7 will never be able to pass a regression test, whatever the compiler/platform. Indeed, in order for this test to even compile, the macro BOOST_UBLAS_USE_INTERVAL needs to be defined.

The Jamfile defines such a macro in the "exe test7" rule, but this rule doesn't seem to be taken into account when regression testing. So the Jamfile of uBLAS should be fixed, or the testing Jamfiles should be enhanced, or...

FWIW, I'm about to finally move uBLAS tests from status/Jamfile to its proper place. I'll take care of this problem along the way.

Done. However, I've added this define only to test7. In theory, it's good to have a set of common defines to apply for all tests, but existing Jamfile seem to declare different set of defines, so I have no idea what to do about it. - Volodya