I think that have solved the remaining issues. The recent changes don't trigger failures in the test suite anymore.
Stephane, can you please go over my commits yesterday and today and check that my changes are OK?
Failed tests in Matlab are 4 block_bytecode/ls2003 tests because of missing fsolve
Failed tests in Octave are 2 conditional_forecasts tests because an Octave graphic bug (test run on Sedna)
Best
Michel
On 09/18/2011 08:33 AM, Sébastien Villemot wrote:
Stéphane Adjemianstepan@dynare.org writes:
I've just found out the problem: nanmean is a routine of the matlab's statistical toolbox! This routine is also distributed in Octave...
I've just pushed a nanmean routine in the utilities/general folder and reverted your last commit (3bf482ae5c3fb0252363d59a6b345b74a6a97e18).
Is there a good reason for not putting 'nanmean' under 'matlab/missing/stats/', like other functions of the MATLAB statistical toolbox that are also in Octave?
@all: I have the impression that matlab never close the generated figures in the test suite... Is this normal ? Shouldn't we close all the figures after each mod file ? On my laptop this seems to be a problem, my matlab freezes after figure number 120...
This is normal but maybe not optimal. We have to see if adding a "close all" after each run does not create problems on text-only terminals.
Best,
Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev