Ticket #12862 (closed defect: fixed)
The partition function under FreeBSD
| Reported by: | stephen | Owned by: | pjeremy |
|---|---|---|---|
| Priority: | major | Milestone: | sage-5.1 |
| Component: | porting: BSD | Keywords: | partition FreeBSD sd40.5 |
| Cc: | Work issues: | ||
| Report Upstream: | N/A | Reviewers: | Mike Hansen |
| Authors: | Stephen Montgomery-Smith | Merged in: | sage-5.1.beta5 |
| Dependencies: | Stopgaps: |
Description
sage built on a FreeBSD 64 bit machine fails the test
sage -t -force_lib "devel/sage/sage/combinat/partition.py"
After some googling, I get the impression that long double is at best flaky on FreeBSD.
I am going to suggest the following patch, although it is perhaps not necessary for the i386 version of FreeBSD.
--- sage-5.0.beta13/sage/combinat/partitions_c.cc-orig 2012-04-20 01:01:41.000000000 +0000 +++ sage-5.0.beta13/sage/combinat/partitions_c.cc 2012-04-20 01:02:47.000000000 +0000 @@ -152,7 +152,7 @@ const unsigned int double_precision = DBL_MANT_DIG; // The assumed precision of a double. -#if defined(__sparc) || defined(__CYGWIN__) +#if defined(__sparc) || defined(__CYGWIN__) || defined(__FreeBSD__) // On sparc solaris long double is bad/broken/different, etc. E.g., // LDBL_MANT_DIG is 113 rather than 106, which causes all kinds of trouble. // So we only use double_precision.
Attachments
Change History
Note: See
TracTickets for help on using
tickets.


Seems like a reasonable change.