On 30 Apr 2011, at 19:35, Thomas Treichl wrote:

> Am 25.04.11 17:29, schrieb SourceForge.net:
>> Bugs item #3292411, was opened at 2011-04-24 11:16
>> Message generated for change (Comment added) made by treichl
>> You can respond by visiting:
>> https://sourceforge.net/tracker/?func=detail&atid=102888&aid=3292411&group_id=2888
>> 
>> Please note that this message will contain a full copy of the comment thread,
>> including the initial issue submission, for this request,
>> not just the latest update.
>> Category: None
>> Group: None
>> Status: Open
>> Resolution: None
>> Priority: 5
>> Private: No
>> Submitted By: Thomas Weber (thomas-weber)
>> Assigned to: Nobody/Anonymous (nobody)
>> Summary: Failing tests in odepkg 0.6.12
>> 
>> Initial Comment:
>> Running the tests in the .cc files of odepkg 0.6.12 with Octave 3.2.4 gives 
>> a lot of output, including some failing tests (the names in []-brackets give 
>> the corresponding .cc file):
>> ==========================================================
>> [odepkg_octsolver_seulex]
>>  EXIT OF SEULEX AT X= 0.2665723D+01   H= 0.4191942D+00
>>  EXIT OF SEULEX AT X= 0.2665723D+01   H= 0.4191942D+00
>> PASSES 32 out of 32 tests
>> [odepkg_octsolver_rodas]
>>  EXIT OF RODAS AT X=        0.2499E+01
>>  EXIT OF RODAS AT X=        0.2499E+01
>> PASSES 32 out of 32 tests
>> [odepkg_octsolver_radau5]
>>  EXIT OF RADAU5 AT X=        0.2496E+01
>>  EXIT OF RADAU5 AT X=        0.2496E+01
>> PASSES 32 out of 32 tests
>> [odepkg_octsolver_radau]
>>  EXIT OF RADAU AT X=        0.2671E+01
>>  EXIT OF RADAU AT X=        0.2671E+01
>> PASSES 32 out of 32 tests
>> [odepkg_octsolver_mebdfi]
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  KFLAG = -3 FROM INTEGRATOR AT T =   0.30000000E-05
>>   CORRECTOR CONVERGENCE COULD NOT BE ACHIEVED
>> 
>> 
>> 
>>  PROBLEM APPEARS UNSOLVABLE WITH GIVEN INPUT
>>          HMIN REDUCED BY A FACTOR OF 1.0E10
>> 
>> 
>>   ***** test %# one output argument
>>   vsol = odebdi (@fpol, [0, 2], [2; 0], [0; -2]);
>>   assert ([vsol.x(end), vsol.y(end,:)], [2, fref], 1e-3);
>>   assert (isfield (vsol, 'solver'));
>>   assert (vsol.solver, 'odebdi');
>> !!!!! test failed
>> Integration was halted after failing to achieve a corrector convergence  
>> even after reducing the step size h by a factor of 1e-10 (error occured in 
>> "mebdfi" core solver function with error number 
>> "-3")[odepkg_octsolver_mebdfdae]
>> PASSES 31 out of 31 tests
>> ==========================================================
>> 
>> Is this output to be expected?
>> 
>> ----------------------------------------------------------------------
>> 
>>> Comment By: Thomas Treichl (treichl)
>> Date: 2011-04-25 15:29
>> 
>> Message:
>> Hi Thomas,
>>   this bug is not expected. I'l take a look at it and report again later.
>> Regards
>>   Thomas
>> 
>> ----------------------------------------------------------------------
> 
> I've taken a look at the problem: The contents of odepkg-0.6.12.tar.gz is
> completely different from odepkg-0.6.10.tar.gz. There are some files that
> shouldn't be there (eg. the .svnignore files) in odepkg-0.6.12.tar.gz and
> some other files that are missing (eg. odepkg.pdf) or that are different
> (eg. Makeconf, Makefile).
> 
> I am a bit out of practice and I used to use the package build
> instructions found at http://octave.sourceforge.net/developers.html
> If I follow these instructions then all these differences appear.
> 
> Now, I used the old traditional way (I still could remember :) that was
> 
> ./configure
> make packages
> cd packages
> make checkpkg/main/odepkg-0.6.12.tar.gz
> 
> and this looks very pretty:
> 
> make checkpkg/main/odepkg-0.6.12.tar.gz
> Building main/odepkg-0.6.12.tar.gz and its dependencies
> Running test code for main/odepkg-0.6.12.tar.gz and its dependencies
> code/odepkg-0.6.12/inst [tests 15 of 35 files]
> ---> success
> code/odepkg-0.6.12/src [tests 7 of 8 files]
> EXIT OF RADAU AT X= 0.2671E+01
> EXIT OF RADAU AT X= 0.2671E+01
> EXIT OF RADAU5 AT X= 0.2496E+01
> EXIT OF RADAU5 AT X= 0.2496E+01
> EXIT OF RODAS AT X= 0.2499E+01
> EXIT OF RODAS AT X= 0.2499E+01
> EXIT OF SEULEX AT X= 0.2665723D+01 H= 0.4191942D+00
> EXIT OF SEULEX AT X= 0.2665723D+01 H= 0.4191942D+00
> ---> success
> passes 610 out of 610 tests
> 
> The exit codes will never disappear. They have already been there in older
> odepkg(s). They come from direct implementations of outputs in the original
> Fortran solver files.
> 
> A detailed look at the mebdfi solver now brings up
> octave-3.2.3:2> test odepkg_octsolver_mebdfi.cc
> PASSES 28 out of 28 tests
> 
> I've uploaded a fixed odepkg-0.6.12.tar.gz file to the package release
> forum at https://sourceforge.net/apps/phpbb/octave/viewforum.php?f=1 under 
> odepkg-0.6.12 in the hope that someone can please replace the broken package 
> from the download site.
> 
> Sorry for the trouble and
> Best regards
> Thomas

I uploaded the new version,
but in the future I think even for such a small change the version number 
should be incremented.
c.
------------------------------------------------------------------------------
WhatsUp Gold - Download Free Network Management Software
The most intuitive, comprehensive, and cost-effective network 
management toolset available today.  Delivers lowest initial 
acquisition cost and overall TCO of any competing solution.
http://p.sf.net/sfu/whatsupgold-sd
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev

Reply via email to