Fri 23 Dec 2016 05:34:03 PM UTC, comment #4:
Final issue, ode23.m and ode45.m have statistics which they print. The new ode15i.m and ode15s.m accept the Stats option, and even have code to set the internal variable options.havestats, but don't print anything.
Should a new bug report be opened about that, or should we just continue to re-purpose this one? Either we can get the statistics out of SUNDIALS in some manner, or we should just issue a warning that no statistics are available.
|
Fri 23 Dec 2016 05:31:21 PM UTC, comment #3:
@Carlo: I committed your cset here (http://hg.savannah.gnu.org/hgweb/octave/rev/bb452f84a299). One thing I did was change the BIST tests to use evalc to capture the output from the Stats option which was otherwise being dumped to stdout after your change.
Also, I don't think there is a reason to follow Matlab slavishly in regards to case sensitivity. When you write something like,
Matlab ignores case and sets the field name 'Stats'. I think Octave should be equally tolerant of the right-hand side argument where people might type 'ON', 'On', or 'on'. I made that additional change here (http://hg.savannah.gnu.org/hgweb/octave/rev/c9344df03da5).
|
Fri 23 Dec 2016 09:10:15 AM UTC, comment #2:
Actually, it seems that in Octave the 'stats' option is neglected
independent of the capitalization ...
I'm changing the title of this ticket and attaching a changeset.
(file #39298)
|
Fri 23 Dec 2016 08:51:23 AM UTC, comment #1:
Although it looks odd, the current behavior seems to be compatible:
|
Fri 23 Dec 2016 06:13:44 AM UTC, original submission:
I don't know how Matlab handles the Stats option to odeset, but Octave behaves in a case-sensitive manner.
But if you look at the code in the m-files, for example ode45.m, you find
It seems like these calls should be replaced with strcmpi. It also looks like the option NormControl requires the same change.
|