[Math] Help needed with FreeBSD failures
Can anyone familiar with FreeBSD help me figure out what's causing a large number of runtime failures from Boost.Math, for example http://www.boost.org/development/tests/develop/developer/output/Flast-FreeBS... Strangely gcc-5.1.0 is fine, but 4.8.5, 5.2.0 and 6.0 all fail. Many thanks in advance, John.
Hi John,
Can anyone familiar with FreeBSD help me figure out what's causing a large number of runtime failures from Boost.Math, for example http://www.boost.org/development/tests/develop/developer/output/Flast-FreeBS...
I'll investigate soon (this is my runner).
Strangely gcc-5.1.0 is fine, but 4.8.5, 5.2.0 and 6.0 all fail.
The col of 5.1.0 is no longer valid because the runner had upgraded recentry 5.1.0 -> 5.2.0. Best, Kohei Takahashi
On 16/10/2015 01:08, Kohei Takahashi wrote:
Hi John,
Can anyone familiar with FreeBSD help me figure out what's causing a large number of runtime failures from Boost.Math, for example http://www.boost.org/development/tests/develop/developer/output/Flast-FreeBS... I'll investigate soon (this is my runner).
Thanks. I've managed to get FreeBSD 10.2 more or less functional locally, and with a default install of gcc-5.2 any test that throws an exception aborts, this causes most of the regex lib tests to fail as well. Changing user-config.jam to add -Wl,-rpath=/usr/local/lib/gcc5 fixes things. Could this be the issue here? Thanks, John.
On 16/10/2015 01:08, Kohei Takahashi wrote:
Hi John,
Can anyone familiar with FreeBSD help me figure out what's causing a large number of runtime failures from Boost.Math, for example http://www.boost.org/development/tests/develop/developer/output/Flast-FreeBS...
I'll investigate soon (this is my runner).
Thanks. I've managed to get FreeBSD 10.2 more or less functional locally, and with a default install of gcc-5.2 any test that throws an exception aborts, this causes most of the regex lib tests to fail as well. Changing user-config.jam to add -Wl,-rpath=/usr/local/lib/gcc5 fixes things. Could this be the issue here? No, it was caused by ABI/API mismatching with ICU. The ICU is linked to
On 2015/10/17 2:36, John Maddock wrote: libc++/libcxxrt what I was installed via pkgng. Now, I removed ICU and tests ran normally. Thanks, Kohei
No, it was caused by ABI/API mismatching with ICU. The ICU is linked to libc++/libcxxrt what I was installed via pkgng. Now, I removed ICU and tests ran normally.
Thanks for tracking that down! I've updated the configure check for ICU that will hopefully detect dll-mismatches now. John.
participants (2)
-
John Maddock -
Kohei Takahashi