On 2009-Aug-10 06:53:38 -0700, William Stein <wst...@gmail.com> wrote: >On Mon, Aug 10, 2009 at 4:55 AM, Peter Jeremy ><peterjer...@optushome.com.au>wrote: > >> (I realise sage-4.1 has been superseded but it seemed less effort to >> complete the porting work on a single release and forward-port the >> patches in one step rather than continuously forward-porting bits. I >> have an equivalent 4.1.1.rc2 patchset in boxen:~peter/sage-4.1.1.rc2.patch >> The "make test" is still running).
The 4.4.1.rc2 "make test" reported the same failures as 4.1. I've updated http://wiki.sagemath.org/freebsd/sage-4.1 to cover 4.1.1.rc2 as well. >> - No csin(), f_cdf() >This requires writing a little code. I thought I saw a csin() during the build so I'll investigate and add code as necessary. >> sage -t "4.1/devel/sage/sage/rings/number_field/number_field.py" >> - Slight numerical differences eg '-0.629960524947437' instead of >> pacomplex'-0.62996052494743...'. > >Change the doctest (put more ...'s). That's the easiest solution. It would be useful to know why there are these numeric differences. I know there are two kludges in the code (using sqrt() instead of sqrtl() and using log2(x) = ln(x) / ln(2) - both of which lose precision but I'm not sure if either is in this calculation path). >> - Whitespace differences > >? Different indenting - again, something to investigate. >> * Incorrect signs on imaginary parts >Are they very small imaginary parts? Unfortunately, no. I will have to investigate both this and the period_lattice.py test failures. Identical numeric parts with an incorrect sign is worrying. >> sage -t "4.1/devel/sage/sage/rings/tests.py" >> * test_random_elements and test_random_arith report "memory usage not >> implemented. > >You just have to implement a "get_memory_usage" function in misc/somewhere. Thanks. I'll look into this. It shouldn't be too difficult. -- Peter Jeremy
pgpQN0KLgLlOv.pgp
Description: PGP signature
