[octave forge] control: support Octave >3.0.0?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

[octave forge] control: support Octave >3.0.0?

Olaf Till-2
Hi Doug,

you seem to have done much work to bring the control package to its
release state now. But do you really want to support Octave versions
down to 3.0.0?  I had bumped it from originally >3.8.0 to >4.0.0 to
make our life easier with the linked code, >3.0.0 would be more
versions than before, and probably too difficult to handle...

Once we have agreed on the supported Octave versions, would you like
me to produce a release tarball?

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Doug Stewart-4


On Wed, Mar 7, 2018 at 4:18 PM, Olaf Till <[hidden email]> wrote:
Hi Doug,

you seem to have done much work to bring the control package to its
release state now. But do you really want to support Octave versions
down to 3.0.0?  I had bumped it from originally >3.8.0 to >4.0.0 to
make our life easier with the linked code, >3.0.0 would be more
versions than before, and probably too difficult to handle...

Once we have agreed on the supported Octave versions, would you like
me to produce a release tarball?

Olaf



I made a mistake, you are correct >4.0.0 is the minimum way to go but...
I have tried it on  4.0.1-rc4 ( which i have on an old machine) and it installs but fails a lot of tests,
so we should  go to >4.1.0

Test on 4.2.0 all pass
Test on stable all pass.
Test on default  some fails that I am looking into.  they segfault octave when calling fortran code.
    I think it is a pointer size problem but not sure yet.

So if you fix this and a new release data then I would suggest that you do a release now. 
There will always be more things to fix and there is time  before default gets released.

thanks for your help.
doug



--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net



--
DASCertificate for 206392



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Olaf Till-2
On Wed, Mar 07, 2018 at 07:27:51PM -0500, Doug Stewart wrote:
> Test on 4.2.0 all pass
> Test on stable all pass.

I get some failures with Octave release 4.2.1:

inst/bstmodred.m: %!assert (Mo, Me, 1e-4); fails due to opposite signs
in some elements of Mo vs. Me

inst/damp.m: First test fails because A is not an "lti" system. The
error message is misleading, seemingly because the input testing in
damp.m is wrong: It accepts a square matrix even if type is not "lti",
but then pole() is not found, because pole.m is only under @lti/ .

inst/ltimodels.m: 12 tests fail, this group of 6:

%!assert (ac, ac_e, 1e-4);
%!assert (ec, ec_e, 1e-4);
%!assert (bc, bc_e, 1e-4);
%!assert (cc, cc_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

seemingly due to opposite signs in some elements, and this group of 6:

%!assert (ao, ao_e, 1e-4);
%!assert (eo, eo_e, 1e-4);
%!assert (bo, bo_e, 1e-4);
%!assert (co, co_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

except for the 4th not due to opposite signs.

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Doug Stewart-4


On Thu, Mar 8, 2018 at 10:44 AM, Olaf Till <[hidden email]> wrote:
On Wed, Mar 07, 2018 at 07:27:51PM -0500, Doug Stewart wrote:
> Test on 4.2.0 all pass
> Test on stable all pass.

I get some failures with Octave release 4.2.1:

inst/bstmodred.m: %!assert (Mo, Me, 1e-4); fails due to opposite signs
in some elements of Mo vs. Me

inst/damp.m: First test fails because A is not an "lti" system. The
error message is misleading, seemingly because the input testing in
damp.m is wrong: It accepts a square matrix even if type is not "lti",
but then pole() is not found, because pole.m is only under @lti/ .

inst/ltimodels.m: 12 tests fail, this group of 6:

%!assert (ac, ac_e, 1e-4);
%!assert (ec, ec_e, 1e-4);
%!assert (bc, bc_e, 1e-4);
%!assert (cc, cc_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

seemingly due to opposite signs in some elements, and this group of 6:

%!assert (ao, ao_e, 1e-4);
%!assert (eo, eo_e, 1e-4);
%!assert (bo, bo_e, 1e-4);
%!assert (co, co_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

except for the 4th not due to opposite signs.

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net




Yes I saw those. but damp works for me on latest stable.
attached is what I see.


--
DAS



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Doug Stewart-4


On Thu, Mar 8, 2018 at 2:55 PM, Doug Stewart <[hidden email]> wrote:


On Thu, Mar 8, 2018 at 10:44 AM, Olaf Till <[hidden email]> wrote:
On Wed, Mar 07, 2018 at 07:27:51PM -0500, Doug Stewart wrote:
> Test on 4.2.0 all pass
> Test on stable all pass.

I get some failures with Octave release 4.2.1:

inst/bstmodred.m: %!assert (Mo, Me, 1e-4); fails due to opposite signs
in some elements of Mo vs. Me

inst/damp.m: First test fails because A is not an "lti" system. The
error message is misleading, seemingly because the input testing in
damp.m is wrong: It accepts a square matrix even if type is not "lti",
but then pole() is not found, because pole.m is only under @lti/ .

inst/ltimodels.m: 12 tests fail, this group of 6:

%!assert (ac, ac_e, 1e-4);
%!assert (ec, ec_e, 1e-4);
%!assert (bc, bc_e, 1e-4);
%!assert (cc, cc_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

seemingly due to opposite signs in some elements, and this group of 6:

%!assert (ao, ao_e, 1e-4);
%!assert (eo, eo_e, 1e-4);
%!assert (bo, bo_e, 1e-4);
%!assert (co, co_e, 1e-4);
%!assert (q, q_e, 1e-4);
%!assert (z, z_e, 1e-4);

except for the 4th not due to opposite signs.

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net




Yes I saw those. but damp works for me on latest stable.
attached is what I see.


--
DAS


fwiw    I see the same 12 fails with control-3.0.0

inst/ltimodels.m: 12 tests fail

so that problem has been here a long time.
I think we can release and I will fix them latter.

--
DASCertificate for 206392



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Mike Miller-4
On Thu, Mar 08, 2018 at 15:07:37 -0500, Doug Stewart wrote:
> fwiw    I see the same 12 fails with control-3.0.0
>
> inst/ltimodels.m: 12 tests fail
>
> so that problem has been here a long time.
> I think we can release and I will fix them latter.

You can have a look at https://savannah.gnu.org/bugs/index.php?45008,
this test failure has been discussed and known for some time. If the
tests fail with OpenBLAS due to a sign difference, but pass when using
the reference BLAS library, then it's #45008.

--
mike


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Doug Stewart-4


On Thu, Mar 8, 2018 at 6:50 PM, Mike Miller <[hidden email]> wrote:
On Thu, Mar 08, 2018 at 15:07:37 -0500, Doug Stewart wrote:
> fwiw    I see the same 12 fails with control-3.0.0
>
> inst/ltimodels.m: 12 tests fail
>
> so that problem has been here a long time.
> I think we can release and I will fix them latter.

You can have a look at https://savannah.gnu.org/bugs/index.php?45008,
this test failure has been discussed and known for some time. If the
tests fail with OpenBLAS due to a sign difference, but pass when using
the reference BLAS library, then it's #45008.

--
mike


---

Thank you Mike.

Olaf, with this info I am more confident that we should go ahead with the release.
I also am inclined to mark these as known errors, with the explanation that mike has given. 
I will repeat this in bug #45008

Doug


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Olaf Till-2
On Sat, Mar 10, 2018 at 11:01:17AM -0500, Doug Stewart wrote:

> On Thu, Mar 8, 2018 at 6:50 PM, Mike Miller <[hidden email]> wrote:
>
> > On Thu, Mar 08, 2018 at 15:07:37 -0500, Doug Stewart wrote:
> > > fwiw    I see the same 12 fails with control-3.0.0
> > >
> > > inst/ltimodels.m: 12 tests fail
> > >
> > > so that problem has been here a long time.
> > > I think we can release and I will fix them latter.
> >
> > You can have a look at https://savannah.gnu.org/bugs/index.php?45008,
> > this test failure has been discussed and known for some time. If the
> > tests fail with OpenBLAS due to a sign difference, but pass when using
> > the reference BLAS library, then it's #45008.
> >
> > --
> > mike
> >
> >
> > ---
>
>
> Thank you Mike.
>
> Olaf, with this info I am more confident that we should go ahead with the
> release.
> I also am inclined to mark these as known errors, with the explanation that
> mike has given.
> I will repeat this in bug #45008
>
> Doug
Thanks for the preparations, release 3.1.0 is now online. Please
remember to announce it, and to pull from the repository.

I've included Octave versions 4.0.. into DESCRIPTION, even if there
are currently more test failures with it than with 4.2.. . 4.0.. is
the Octave version provided by Debian stable, so if possible we should
support it. imp_invar and damp had to be added to INDEX, hopefully
they are in the correct section.

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net


-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [octave forge] control: support Octave >3.0.0?

Doug Stewart-4


On Sun, Mar 11, 2018 at 1:00 PM, Olaf Till <[hidden email]> wrote:
On Sat, Mar 10, 2018 at 11:01:17AM -0500, Doug Stewart wrote:
> On Thu, Mar 8, 2018 at 6:50 PM, Mike Miller <[hidden email]> wrote:
>
> > On Thu, Mar 08, 2018 at 15:07:37 -0500, Doug Stewart wrote:
> > > fwiw    I see the same 12 fails with control-3.0.0
> > >
> > > inst/ltimodels.m: 12 tests fail
> > >
> > > so that problem has been here a long time.
> > > I think we can release and I will fix them latter.
> >
> > You can have a look at https://savannah.gnu.org/bugs/index.php?45008,
> > this test failure has been discussed and known for some time. If the
> > tests fail with OpenBLAS due to a sign difference, but pass when using
> > the reference BLAS library, then it's #45008.
> >
> > --
> > mike
> >
> >
> > ---
>
>
> Thank you Mike.
>
> Olaf, with this info I am more confident that we should go ahead with the
> release.
> I also am inclined to mark these as known errors, with the explanation that
> mike has given.
> I will repeat this in bug #45008
>
> Doug

Thanks for the preparations, release 3.1.0 is now online. Please
remember to announce it, and to pull from the repository.

I've included Octave versions 4.0.. into DESCRIPTION, even if there
are currently more test failures with it than with 4.2.. . 4.0.. is
the Octave version provided by Debian stable, so if possible we should
support it. imp_invar and damp had to be added to INDEX, hopefully
they are in the correct section.

Olaf

--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

Thank you Olaf.



--
DAS



-----------------------------------------
Join us March 12-15 at CERN near Geneva
Switzerland for OctConf 2018.  More info:
https://wiki.octave.org/OctConf_2018
-----------------------------------------