marco atzeri-2 wrote
that test was always problematic
https://mailman.cae.wisc.edu/pipermail/octave-maintainers/2011-August/024469.html
the solution is found with iteration from a random start point
and different system/compiler/library could need a
different start point.
I suggest to skip it as it should need a math rethink;
otherwise you could test several start point and
confirm the failure if all fails
Regards
Marco
noob mistake on my part. i did not even think to check the mailing list
first. i wil do so in the future.
I followed test suggested on Aug 2011 thread (try various starting points).
I found no successful pass with seed in the range [35,50]. Im on
Linux.
Given how finicky this test is, and that failures are being ignored anyhow
is there any value in keeping this particular test? Would anyone be
distressed if i submit a patch that removes it?
A test that fails and is habitually ignored would seem (to me) to be of no
value. If tests are to be taken seriously, they need to work. In this
particular case, the lack of comments to state the intent of the test is
also not helpful. It should be possible to state the purpose of the test
so someone who is not a domain expert can understand the intent.