[Octave Forge] Octave 4.0 call for packages

classic Classic list List threaded Threaded
80 messages Options
1234
Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

tmacchant




----- Original Message -----

> From: Julien Bect 
> To: octave-maintainers
> Cc:
> Date: 2015/4/8, Wed 14:02
> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>
> Le 08/04/2015 04:08, Tatsuro MATSUOKA a écrit :
>>  I think that it is worth trying made a new package for octave-4.0.0.
>>
>>  Please point me instruction with which one can make release package in
> mercurial repository.
>
> http://octave.sourceforge.net/developers.html


Thanks!

Tatsuro

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Daniel Kraft-2
In reply to this post by Philip Nienhuis
Hi!

On 2015-04-07 20:47, Philip Nienhuis wrote:

> Mike Miller-4 wrote
>> The same logic as above applies, they still exist in
>> specfun because no one has made a new release of the package to remove
>> those functions since being added to core. I mentioned earlier in
>> another thread that I would be in favor of dropping the entire specfun
>> package now.
>
> The specfun package contains several functions that appear to be useful.
> I'm not against a new release w/o the shadowing functions as long as the
> rest isn't harmful.
Exactly -- I, for instance, have specfun installed for "lambertw".  I
needed it already for two (different) things.  As far as I can tell, it
is not available in core yet.

Unless *all* those functions are ported to core, I think the specfun
package should remain.

Yours,
Daniel

--
http://www.domob.eu/
OpenPGP: 1142 850E 6DFF 65BA 63D6  88A8 B249 2AC4 A733 0737
Namecoin: id/domob -> https://nameid.org/?name=domob
--
Done:  Arc-Bar-Cav-Hea-Kni-Ran-Rog-Sam-Tou-Val-Wiz
To go: Mon-Pri


signature.asc (853 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

jcorno
In reply to this post by c.-2


Il 07/04/2015 20:12, Carlo de Falco ha scritto:

odepkg does have a maintener but its development moved out of sf into a separate repo.
Jacopo can you comment on the status of odepkg in general and odepkg in particular?
c.


The repository Carlo is mentioning is

https://bitbucket.org/robs88/octave-odepkg

The structure that was used in odepkg-0.8.4 was changed two years ago for SoCiS together with the introduction of geometric integrators. Some more info on the current version is here (http://geointegratorssocis.blogspot.de/)

More recently I cleaned up the code and modified it to allow all Matlab options to be used at least for the most commonly used solvers like ode45. The plan is to push these solvers to the core, but I still have to take care of adding those to the tracker.

Best,
Jacopo
Il 07/apr/2015 13:21 "Carnë Draug" <[hidden email]> ha scritto:
On 7 April 2015 at 11:30, tmacchant <[hidden email]> wrote:
> [...]
> Now control and image packages work on 4.0. Right?

Yes.

> I use the odepkg package in order to run Matlab code as it is.
> (ode45)
>
> If  odepkg does not work on 4.0, I have to write rapper to lsode or the code
> is modified to use lsode instead.
>
> Does anyone know of the state of the odepkg package?

The version on the repositories should work on 4.0.0.  But the package does
not have a maintainer that will clean it up and make a release.  So you could
use it but you'll have to prepare the package yourself.

Carnë


Reply | Threaded
Open this post in threaded view
|

Re: Re: [Octave Forge] Octave 4.0 call for packages

tmacchant
In reply to this post by tmacchant
--- Jacopo Corno < wrote:

> Il 07/04/2015 20:12, Carlo de Falco ha
>       scritto:
>
>  odepkg does have a maintener but its development
>         moved out of sf into a separate repo.
> Jacopo can you comment on the status of odepkg in general and
>         odepkg in particular?
> c.
>    
>    
> The repository Carlo is mentioning is
>
> https://bitbucket.org/robs88/octave-odepkg
>
> The structure that was used in odepkg-0.8.4 was changed two years
>     ago for SoCiS together with the introduction of geometric
>     integrators. Some more info on the current version is here
>     (http://geointegratorssocis.blogspot.de/)
>
> More recently I cleaned up the code and modified it to allow all
>     Matlab options to be used at least for the most commonly used
>     solvers like ode45. The plan is to push these solvers to the core,
>     but I still have to take care of adding those to the tracker.
>
> Best,
> Jacopo
>
>      
> Il 07/apr/2015 13:21 "Carnë Draug"
>         ha scritto:
> On 7 April
>           2015 at 11:30, tmacchant  wrote:
> > [...]
> > Now control and image packages work on 4.0. Right?
>
> Yes.
>
> > I use the odepkg package in order to run Matlab code as
>           it is.
> > (ode45)
> >
> > If  odepkg does not work on 4.0, I have to write rapper
>           to lsode or the code
> > is modified to use lsode instead.
> >
> > Does anyone know of the state of the odepkg package?
>
> The version on the repositories should work on 4.0.0.  But the
>           package does
> not have a maintainer that will clean it up and make a
>           release.  So you could
> use it but you&#39;ll have to prepare the package yourself.
>
> Carnë
>
I am very glad to hear that odepkg functions are revised on the sparated project.  However, if the support of the odepkg package is stopped at 4.0.0 release, many people will be disappointed.

The odepkg package on the mercurial repository can be compiled on octave 4.0.0-rc2 with small patch to src/Makefile by JohnD.

Please allow me to push the temporal release of the odepkg package on the current mercurial repository for upcoming octave release of 4.0.0.

http://savannah.gnu.org/bugs/?44769

Tatsuro

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

c.-2

On 8 Apr 2015, at 15:56, Tatsuro MATSUOKA <[hidden email]> wrote:

> --- Jacopo Corno < wrote:
>> Il 07/04/2015 20:12, Carlo de Falco ha
>>      scritto:
>>
>> odepkg does have a maintener but its development
>>        moved out of sf into a separate repo.
>> Jacopo can you comment on the status of odepkg in general and
>>        odepkg in particular?
>> c.
>>
>>
>> The repository Carlo is mentioning is
>>
>> https://bitbucket.org/robs88/octave-odepkg
>>
>> The structure that was used in odepkg-0.8.4 was changed two years
>>    ago for SoCiS together with the introduction of geometric
>>    integrators. Some more info on the current version is here
>>    (http://geointegratorssocis.blogspot.de/)
>>
>> More recently I cleaned up the code and modified it to allow all
>>    Matlab options to be used at least for the most commonly used
>>    solvers like ode45. The plan is to push these solvers to the core,
>>    but I still have to take care of adding those to the tracker.
>>
>> Best,
>> Jacopo
>>
>>
>> Il 07/apr/2015 13:21 "Carnë Draug"
>>        ha scritto:
>> On 7 April
>>          2015 at 11:30, tmacchant  wrote:
>>> [...]
>>> Now control and image packages work on 4.0. Right?
>>
>> Yes.
>>
>>> I use the odepkg package in order to run Matlab code as
>>          it is.
>>> (ode45)
>>>
>>> If  odepkg does not work on 4.0, I have to write rapper
>>          to lsode or the code
>>> is modified to use lsode instead.
>>>
>>> Does anyone know of the state of the odepkg package?
>>
>> The version on the repositories should work on 4.0.0.  But the
>>          package does
>> not have a maintainer that will clean it up and make a
>>          release.  So you could
>> use it but you&#39;ll have to prepare the package yourself.
>>
>> Carnë
>>
> I am very glad to hear that odepkg functions are revised on the sparated project.  However, if the support of the odepkg package is stopped at 4.0.0 release, many people will be disappointed.
>
> The odepkg package on the mercurial repository can be compiled on octave 4.0.0-rc2 with small patch to src/Makefile by JohnD.
>
> Please allow me to push the temporal release of the odepkg package on the current mercurial repository for upcoming octave release of 4.0.0.
>
> http://savannah.gnu.org/bugs/?44769
>
> Tatsuro


Tatsuro,

The new repository is NOT a separate project,
that is just the current development version
of the package, the one on sourceforge has not
been yet updated to include latest changes.

This is, to my understanding, the new trend in
Forge packages, development happens elsewhere then
the repository is updated just before the release.

The sourceforge repository was not updated because
no release was made yet. I believe it would anyway
make more sense to synchronize the two repos now to
avoid more confusion, so if you want to merge the
changes on bitbucket to sourceforge, please go ahead.

BTW do the changes made to allow compatibility with
4.0 break 3.8 compatibility?

Many people (e.g. Mac users) will be stuck with Octave 3.8
for a while, so if

pkg install -forge odepkg

will not work for them that may be a problem...

c.



Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

tmacchant
----- Original Message -----

> From: c. 
> To: Tatsuro MATSUOKA 
> Cc: Carnë Draug ; Sebastian Schöps ; Jacopo Corno ; octave-maintainers
> Date: 2015/4/9, Thu 01:47
> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>
>
> On 8 Apr 2015, at 15:56, Tatsuro MATSUOKA <[hidden email]> wrote:
>
>>  --- Jacopo Corno < wrote:
>>>  Il 07/04/2015 20:12, Carlo de Falco ha
>>>       scritto:
>>>
>>>  odepkg does have a maintener but its development
>>>         moved out of sf into a separate repo.
>>>  Jacopo can you comment on the status of odepkg in general and
>>>         odepkg in particular?
>>>  c.
>>>
>>>
>>>  The repository Carlo is mentioning is
>>>
>>>  https://bitbucket.org/robs88/octave-odepkg
>>>
>>>  The structure that was used in odepkg-0.8.4 was changed two years
>>>     ago for SoCiS together with the introduction of geometric
>>>     integrators. Some more info on the current version is here
>>>     (http://geointegratorssocis.blogspot.de/)
>>>
>>>  More recently I cleaned up the code and modified it to allow all
>>>     Matlab options to be used at least for the most commonly used
>>>     solvers like ode45. The plan is to push these solvers to the core,
>>>     but I still have to take care of adding those to the tracker.
>>>
>>>  Best,
>>>  Jacopo
>>>
>>>
>>>  Il 07/apr/2015 13:21 "Carnë Draug"
>>>         ha scritto:
>>>  On 7 April
>>>           2015 at 11:30, tmacchant  wrote:
>>>>  [...]
>>>>  Now control and image packages work on 4.0. Right?
>>>
>>>  Yes.
>>>
>>>>  I use the odepkg package in order to run Matlab code as
>>>           it is.
>>>>  (ode45)
>>>>
>>>>  If  odepkg does not work on 4.0, I have to write rapper
>>>           to lsode or the code
>>>>  is modified to use lsode instead.
>>>>
>>>>  Does anyone know of the state of the odepkg package?
>>>
>>>  The version on the repositories should work on 4.0.0.  But the
>>>           package does
>>>  not have a maintainer that will clean it up and make a
>>>           release.  So you could
>>>  use it but you'll have to prepare the package yourself.
>>>
>>>  Carnë
>>>
>>  I am very glad to hear that odepkg functions are revised on the sparated
> project.  However, if the support of the odepkg package is stopped at 4.0.0
> release, many people will be disappointed.
>>
>>  The odepkg package on the mercurial repository can be compiled on octave
> 4.0.0-rc2 with small patch to src/Makefile by JohnD.
>>
>>  Please allow me to push the temporal release of the odepkg package on the
> current mercurial repository for upcoming octave release of 4.0.0.
>>
>>  http://savannah.gnu.org/bugs/?44769
>>
>>  Tatsuro
>
>
> Tatsuro,
>
> The new repository is NOT a separate project,
> that is just the current development version
> of the package, the one on sourceforge has not
> been yet updated to include latest changes.
>
> This is, to my understanding, the new trend in
> Forge packages, development happens elsewhere then
> the repository is updated just before the release.
>
> The sourceforge repository was not updated because
> no release was made yet. I believe it would anyway
> make more sense to synchronize the two repos now to
> avoid more confusion, so if you want to merge the
> changes on bitbucket to sourceforge, please go ahead.
>
> BTW do the changes made to allow compatibility with
> 4.0 break 3.8 compatibility?
>
> Many people (e.g. Mac users) will be stuck with Octave 3.8
> for a while, so if
>
> pkg install -forge odepkg
>
> will not work for them that may be a problem...
>
> c.



> The new repository is NOT a separate project, 
> that is just the current development version 
> of the package, the one on sourceforge has not
> been yet updated to include latest changes.


I understand.

> BTW do the changes made to allow compatibility with
> 4.0 break 3.8 compatibility?

> Many people (e.g. Mac users) will be stuck with Octave 3.8
> for a while, so if 

> pkg install -forge odepkg 

> will not work for them that may be a problem...


I have confirmed the 3.8 compatibility in current mercurial repository source on ubuntu 14.04 (32 and 64 bit) and window 7 (32 bit).


> The sourceforge repository was not updated because
> no release was made yet. I believe it would anyway
> make more sense to synchronize the two repos now to 
> avoid more confusion, so if you want to merge the
> changes on bitbucket to sourceforge, please go ahead.


Hmmm. 

I do not have plan to do so.
What is important is now that peoples can use odepkg package on upcoming release ver. 4.0.0.

I will switch my intention to make the patched tar-ball of the odepkg-0.8.4 (e.g. octave-0.8.4_oct4.tar.gz) but not odepkg-0.8.5. The patched tar-ball will be uploaded my personal web site and I announce it on octave-mailing list.

Tatsuro 

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

tmacchant
----- Original Message -----

> From: Tatsuro MATSUOKA
> To: Carlo de Falco 
> Cc: Jacopo Corno; "octave-maintainers; Sebastian Schöps 
> Date: 2015/4/9, Thu 08:33
> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>
> ----- Original Message -----
>
>>  From: c. 
>>  To: Tatsuro MATSUOKA 
>>  Cc: Carnë Draug ; Sebastian Schöps ; Jacopo Corno ; octave-maintainers
>>  Date: 2015/4/9, Thu 01:47
>>  Subject: Re: [Octave Forge] Octave 4.0 call for packages
>>
>>
>>  On 8 Apr 2015, at 15:56, Tatsuro MATSUOKA <[hidden email]>
> wrote:
>>
>>>   --- Jacopo Corno < wrote:
>>>>   Il 07/04/2015 20:12, Carlo de Falco ha
>>>>        scritto:
>>>>
>>>>   odepkg does have a maintener but its development
>>>>          moved out of sf into a separate repo.
>>>>   Jacopo can you comment on the status of odepkg in general and
>>>>          odepkg in particular?
>>>>   c.
>>>>
>>>>
>>>>   The repository Carlo is mentioning is
>>>>
>>>>   https://bitbucket.org/robs88/octave-odepkg
>>>>
>>>>   The structure that was used in odepkg-0.8.4 was changed two years
>>>>      ago for SoCiS together with the introduction of geometric
>>>>      integrators. Some more info on the current version is here
>>>>      (http://geointegratorssocis.blogspot.de/)
>>>>
>>>>   More recently I cleaned up the code and modified it to allow all
>>>>      Matlab options to be used at least for the most commonly used
>>>>      solvers like ode45. The plan is to push these solvers to the
> core,
>>>>      but I still have to take care of adding those to the tracker.
>>>>
>>>>   Best,
>>>>   Jacopo
>>>>
>>>>
>>>>   Il 07/apr/2015 13:21 "Carnë Draug"
>>>>          ha scritto:
>>>>   On 7 April
>>>>            2015 at 11:30, tmacchant  wrote:
>>>>>   [...]
>>>>>   Now control and image packages work on 4.0. Right?
>>>>
>>>>   Yes.
>>>>
>>>>>   I use the odepkg package in order to run Matlab code as
>>>>            it is.
>>>>>   (ode45)
>>>>>
>>>>>   If  odepkg does not work on 4.0, I have to write rapper
>>>>            to lsode or the code
>>>>>   is modified to use lsode instead.
>>>>>
>>>>>   Does anyone know of the state of the odepkg package?
>>>>
>>>>   The version on the repositories should work on 4.0.0.  But the
>>>>            package does
>>>>   not have a maintainer that will clean it up and make a
>>>>            release.  So you could
>>>>   use it but you'll have to prepare the package yourself.
>>>>
>>>>   Carnë
>>>>
>>>   I am very glad to hear that odepkg functions are revised on the
> sparated
>>  project.  However, if the support of the odepkg package is stopped at 4.0.0
>
>>  release, many people will be disappointed.
>>>
>>>   The odepkg package on the mercurial repository can be compiled on
> octave
>>  4.0.0-rc2 with small patch to src/Makefile by JohnD.
>>>
>>>   Please allow me to push the temporal release of the odepkg package on
> the
>>  current mercurial repository for upcoming octave release of 4.0.0.
>>>
>>>   http://savannah.gnu.org/bugs/?44769
>>>
>>>   Tatsuro
>>
>>
>>  Tatsuro,
>>
>>  The new repository is NOT a separate project,
>>  that is just the current development version
>>  of the package, the one on sourceforge has not
>>  been yet updated to include latest changes.
>>
>>  This is, to my understanding, the new trend in
>>  Forge packages, development happens elsewhere then
>>  the repository is updated just before the release.
>>
>>  The sourceforge repository was not updated because
>>  no release was made yet. I believe it would anyway
>>  make more sense to synchronize the two repos now to
>>  avoid more confusion, so if you want to merge the
>>  changes on bitbucket to sourceforge, please go ahead.
>>
>>  BTW do the changes made to allow compatibility with
>>  4.0 break 3.8 compatibility?
>>
>>  Many people (e.g. Mac users) will be stuck with Octave 3.8
>>  for a while, so if
>>
>>  pkg install -forge odepkg
>>
>>  will not work for them that may be a problem...
>>
>>  c.
>
>
>
>>  The new repository is NOT a separate project, 
>>  that is just the current development version 
>>  of the package, the one on sourceforge has not
>>  been yet updated to include latest changes.
>
>
> I understand.
>
>>  BTW do the changes made to allow compatibility with
>>  4.0 break 3.8 compatibility?
>
>>  Many people (e.g. Mac users) will be stuck with Octave 3.8
>>  for a while, so if 
>>  
>>  pkg install -forge odepkg 
>>  
>>  will not work for them that may be a problem...
>
>
> I have confirmed the 3.8 compatibility in current mercurial repository source on
> ubuntu 14.04 (32 and 64 bit) and window 7 (32 bit).
>
>
>>  The sourceforge repository was not updated because
>>  no release was made yet. I believe it would anyway
>>  make more sense to synchronize the two repos now to 
>>  avoid more confusion, so if you want to merge the
>>  changes on bitbucket to sourceforge, please go ahead.
>
>
> Hmmm. 
>
> I do not have plan to do so.
> What is important is now that peoples can use odepkg package on upcoming release
> ver. 4.0.0.
>
> I will switch my intention to make the patched tar-ball of the odepkg-0.8.4
> (e.g. octave-0.8.4_oct4.tar.gz) but not odepkg-0.8.5. The patched tar-ball will
> be uploaded my personal web site and I announce it on octave-mailing list.
>
> Tatsuro 


Apart from the above discussion, the patch by John Donoghue should be applied into the current odepkg mercurial repository as soon as possible.

http://savannah.gnu.org/bugs/?44594  ( file #33856. ).


I do not have write access to the octave-forge repository so that I will ask to those who has write access to the octave-forge repository.

Tatsuro

Reply | Threaded
Open this post in threaded view
|

RE: [Octave Forge] Octave 4.0 call for packages

JohnD
In reply to this post by Carnë Draug


> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf Of
> Carnë Draug
> Sent: Tuesday, April 07, 2015 7:13 AM
> To: John Donoghue
> Cc: octave-maintainers; Mike Miller
> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>
> On 6 April 2015 at 23:31, John Donoghue <[hidden email]>
> wrote:
> > On 04/06/2015 06:16 PM, [hidden email] wrote:
> >>
> >> Message: 4
> >> Date: Mon, 6 Apr 2015 21:17:38 +0100
> >> From: Carn? Draug<[hidden email]>
> >> To: Mike Miller<[hidden email]>
> >> Cc: Octave Maintainers<[hidden email]>
> >> Subject: Re: [Octave Forge] Octave 4.0 call for packages
> >> Message-ID:
> >>
> >> <CAPOrs_1RNATmWYF9unM6NW7QDUxtJT24G-
> [hidden email]>
> >> Content-Type: text/plain; charset=UTF-8
> >>
> >> On 15 March 2015 at 15:04, Mike Miller<[hidden email]>  wrote:
> >>>
> >>> >[...]
> >>> >
> >>> >The imminent Octave 4.0.0 release will be the first to supply an
> >>> >official binary distribution of Octave for Windows operating systems.
> >>> >This distribution will include a set of Octave Forge packages that
> >>> >may be installed at the user's option. This is why we would like to
> >>> >have compatible packages ready to incorporate into the next release
> >>> >candidate for further testing.
> >>> >
> >>
> >> Can users that have installed Octave with such installer then install
> >> more packages and other versions of those packages?
> >>
> >> If we are starting to distribute binaries of Octave core, would be
> >> nice if we can also distribute binaries for the packages, specially
> >> since the release cycle of core and OF is quite different.
> >>
> >> Carn?

I just tried installing all of packages that currently in the active list on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
This is trying to install on the windows machine, not attempting cross compile, using what is available as the 'current' package on octave-forge.
Heres the following results:

bim  1.1.5   - OK
cgi  0.1.0 - OK
communications  1.2.1 - OK - warns unusable help found
control 2.8.0 - OK
database  2.3.1 - OK
dataframe  1.1.0 - OK
data-smoothing  1.3.0 - OK
dicom   0.1.1 - FAILED - Error with octave_map - Bug #44624
divand  1.1.2 - OK
econometrics 1.1.1 - OK
fem-fenics  0.0.3  - FAILED - requires eigen3 lib in mxe-octave, so currently an expected failure
financial 0.4.0 - OK
fits 1.0.5   - OK
fl-core 1.0.0 - FAILED - has been patched in mxe-octave - Bug #41214
fpl  1.3.4 - OK
fuzzy-logic-toolkit 0.4.4 - OK - warning about continuation markers depreciated
ga 0.10.0 - OK - lots of warning about continuation markers depreciated
general  1.3.4 - OK
generate_html 0.1.7 - OK
geometry 1.7.0 - OK - warning on printf args - warn on cant cleanup after self
image   2.4.0 - OK
image-acquisition 0.2.1 - FAILED -  requires v4l library no in mxe-octave so expected
instrument-control 0.2.1 - OK
interval  0.1.5 - OK
io 2.2.7 - OK
java   - part of core now not tried
level-set  0.2.0 - OK
linear-algebra  2.2.2 -OK - warnings on abstract class types
lssa  0.1.2 - OK
ltfat  2.0.1 - FAILED   - 'cc' not found - patched in mxe. - bug #43603
mechanics 1.3.1 - FAILED - install failure  Bug #44804
miscellaneous 1.2.1 - OK
mpi 1.2.0 - FAILED - mxe has no mpi lib, so expected
msh 1.0.10 - OK
mvn 1.1.0 - OK
nan 2.7.1 - OK - warning -fPIC, shadow core funcs
ncarray  1.0.3 - OK
netcdf 1.0.6 - OK
nurbs 1.3.9 - FAILED - matrix_value - bug #44651
ocs  0.1.3 - OK - continuation marker warnings
octcdf 1.1.8 - OK
octclip 1.0.3 - OK - printf format warnings
octproj 1.1.2 - FAILED - mxe-octave needs libproj, so expected failure
odepkg  0.8.4 - FAILED -  octave_map - bug #44594  
optics 0.1.1 - DONE  
optim 1.4.1 - DONE
optiminterp 0.3.4 - DONE
parallel 2.2.1 - FAILED - bug #41148
quaternion 2.4.0 - OK
queueing 1.2.3 - OK
secs1d 0.0.9 - OK
secs2d 0.0.8 - FAILED - Octave_map error Bug #44803
secs3d 0.0.1 - OK
signal 1.3.1 - OK
sockets 1.2.0 - OK
specfun 1.1.0 - OK - warnings on some shadows funcs
splines 1.2.7 - OK
statistics 1.2.4 - OK
stk 2.2.0 - OK
strings 1.1.0 - FAILED - pcre - Bug #44640
struct 1.0.11 - OK
symbolic 2.1.0 - OK
tsa  4.2.9 - OK - warn on shadows a core lib function
vrml 1.0.13 - OK - warnings on continuation markers
windows 1.2.1 - FAILED - Array compile errors - Bug #44180




Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Oliver Heimlich


On 10.04.2015 22:05, JohnD wrote:

>
>
>> -----Original Message-----
>> From: [hidden email] [mailto:[hidden email]] On Behalf Of
>> Carnë Draug
>> Sent: Tuesday, April 07, 2015 7:13 AM
>> To: John Donoghue
>> Cc: octave-maintainers; Mike Miller
>> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>>
>> On 6 April 2015 at 23:31, John Donoghue <[hidden email]>
>> wrote:
>>> On 04/06/2015 06:16 PM, [hidden email] wrote:
>>>>
>>>> Message: 4
>>>> Date: Mon, 6 Apr 2015 21:17:38 +0100
>>>> From: Carn? Draug<[hidden email]>
>>>> To: Mike Miller<[hidden email]>
>>>> Cc: Octave Maintainers<[hidden email]>
>>>> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>>>> Message-ID:
>>>>
>>>> <CAPOrs_1RNATmWYF9unM6NW7QDUxtJT24G-
>> [hidden email]>
>>>> Content-Type: text/plain; charset=UTF-8
>>>>
>>>> On 15 March 2015 at 15:04, Mike Miller<[hidden email]>  wrote:
>>>>>
>>>>>> [...]
>>>>>>
>>>>>> The imminent Octave 4.0.0 release will be the first to supply an
>>>>>> official binary distribution of Octave for Windows operating systems.
>>>>>> This distribution will include a set of Octave Forge packages that
>>>>>> may be installed at the user's option. This is why we would like to
>>>>>> have compatible packages ready to incorporate into the next release
>>>>>> candidate for further testing.
>>>>>>
>>>>
>>>> Can users that have installed Octave with such installer then install
>>>> more packages and other versions of those packages?
>>>>
>>>> If we are starting to distribute binaries of Octave core, would be
>>>> nice if we can also distribute binaries for the packages, specially
>>>> since the release cycle of core and OF is quite different.
>>>>
>>>> Carn?
>
> I just tried installing all of packages that currently in the active list on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
> This is trying to install on the windows machine, not attempting cross compile, using what is available as the 'current' package on octave-forge.
> Heres the following results:

I can confirm your observations. I did an installation on Win 7 64bit
and additionally did some “pkg test”ing. Unfortunately only few packages
contain sufficient tests and allow an assessment. Here are my results
(didn't check them against known bugs):

> bim  1.1.5   - OK
PASS 92, FAIL 0
15 (of 47) .m files have no tests.
> cgi  0.1.0 - OK
6 (of 6) .m files have no tests.
> communications  1.2.1 - OK - warns unusable help found
PASS 457, FAIL 0
19 (of 115) .m files have no tests.
> control 2.8.0 - OK
PASS 317, FAIL 13
229 (of 287) .m files have no tests.
> database  2.3.1 - OK
PASS 0, FAIL 1 (should be expected)
5 (of 6) .m files have no tests.
> dataframe  1.1.0 - OK
PASS 3, FAIL 9
101 (of 101) .m files have no tests.
> data-smoothing  1.3.0 - OK
4 (of 4) .m files have no tests.
> dicom   0.1.1 - FAILED - Error with octave_map - Bug #44624
-
> divand  1.1.2 - OK
PASS 1, FAIL 0
115 (of 116) .m files have no tests.
> econometrics 1.1.1 - OK
28 (of 28) .m files have no tests.
> fem-fenics  0.0.3  - FAILED - requires eigen3 lib in mxe-octave, so currently an expected failure
-
> financial 0.4.0 - OK
PASS 158, FAIL 9
23 (of 57) .m files have no tests.
> fits 1.0.5   - OK
No m-files for testing.
> fl-core 1.0.0 - FAILED - has been patched in mxe-octave - Bug #41214
-
> fpl  1.3.4 - OK
PASS 0, FAIL 5 (failed calls to deprecated function files)
10 (of 10) .m files have no tests.
> fuzzy-logic-toolkit 0.4.4 - OK - warning about continuation markers depreciated
I did not see a warning?!
50 (of 50) .m files have no tests.
> ga 0.10.0 - OK - lots of warning about continuation markers depreciated
I did not see a warning?!
PASS 79, FAIL 0
16 (of 22) .m files have no tests.
> general  1.3.4 - OK
PASS 25, FAIL 0, XFAIL 1
24 (of 28) .m files have no tests.
> generate_html 0.1.7 - OK
8 (of 8) .m files have no tests.
> geometry 1.7.0 - OK - warning on printf args - warn on cant cleanup after self
I can confirm the printf warnings only.
PASS 282, FAIL 3 (missing python, should be expected)
188 (of 233) .m files have no tests.
> image   2.4.0 - OK
PASS 631, FAIL 5, XFAIL 21
54 (of 137) .m files have no tests.
> image-acquisition 0.2.1 - FAILED -  requires v4l library no in mxe-octave so expected
-
> instrument-control 0.2.1 - OK
27 (of 27) .m files have no tests
> interval  0.1.5 - OK
PASS 7264, FAIL 0, XFAIL 18
0 (of 249) .m files have no tests.
> io 2.2.7 - OK
PASS 11, FAIL 0
30 (of 32) .m files have no tests.
> java   - part of core now not tried
-
> level-set  0.2.0 - OK
Compile Warning: -fPIC ignored for target
PASS 198, FAIL 6 ('ls_copy_sign' undefined errors)
2 (of 31) .m files have no tests.
> linear-algebra  2.2.2 -OK - warnings on abstract class types
Several warnings during compile:  writing file 'C:\.octave_hist':
Permission denied
PASS 19, FAIL 1 (function 'fsin' not found)
57 (of 69) .m files have no tests.
> lssa  0.1.2 - OK
PASS 16, FAIL 0
0 (of 7) .m files have no tests.
> ltfat  2.0.1 - FAILED   - 'cc' not found - patched in mxe. - bug #43603
-
> mechanics 1.3.1 - FAILED - install failure  Bug #44804
-
> miscellaneous 1.2.1 - OK
PASS 384, FAIL 9 (strsplit errors)
16 (of 30) .m files have no tests.
> mpi 1.2.0 - FAILED - mxe has no mpi lib, so expected
-
> msh 1.0.10 - OK
PASS 28, FAIL 1 (the gmesh subprocess exited abnormally)
9 (of 21) .m files have no tests.
> mvn 1.1.0 - OK
PASS 1, FAIL 0
18 (of 19) .m files have no tests.
> nan 2.7.1 - OK - warning -fPIC, shadow core funcs
Warnings during Compile:
ls: /usr/local/: No such file or directory
ls: /bin/mexext: No such file or directory
PASS 31, FAIL 0
65 (of 81) .m files have no tests.
> ncarray  1.0.3 - OK
45 (of 45) .m files have no tests.
> netcdf 1.0.6 - OK
Several warnings during compile: error: writing file 'C:\.octave_hist':
Permission denied
10 (of 10) .m files have no tests.
> nurbs 1.3.9 - FAILED - matrix_value - bug #44651
-
> ocs  0.1.3 - OK - continuation marker warnings
Can't install, depends on odepkg
> octcdf 1.1.8 - OK
Several warnings during compile: error: writing file 'C:\.octave_hist':
Permission denied
PASS 23, FAIL 0
30 (of 31) .m files have no tests.
> octclip 1.0.3 - OK - printf format warnings
PASS 5, FAIL 0
0 (of 1) .m files have no tests.
> octproj 1.1.2 - FAILED - mxe-octave needs libproj, so expected failure
-
> odepkg  0.8.4 - FAILED -  octave_map - bug #44594
-
> optics 0.1.1 - DONE
PASS 166, FAIL 2 (%!test should be %!error)
0 (of 43) .m files have no tests.
> optim 1.4.1 - DONE
PASS 16, FAIL 0
49 (of 54) .m files have no tests.
> optiminterp 0.3.4 - DONE
Several warnings during compile: error: writing file 'C:\.octave_hist':
Permission denied
9 (of 9) .m files have no tests.
> parallel 2.2.1 - FAILED - bug #41148
-
> quaternion 2.4.0 - OK
PASS 23, FAIL 0
60 (of 76) .m files have no tests.
> queueing 1.2.3 - OK
PASS 261, FAIL 4 ('sumexpn' undefined, and low tolerances), XFAIL 3
23 (of 101) .m files have no tests.
> secs1d 0.0.9 - OK
5 (of 5) .m files have no tests.
> secs2d 0.0.8 - FAILED - Octave_map error Bug #44803
-
> secs3d 0.0.1 - OK
39 (of 39) .m files have no tests.
> signal 1.3.1 - OK
PASS 469, FAIL 3
78 (of 146) .m files have no tests.
> sockets 1.2.0 - OK
No m-files for testing
> specfun 1.1.0 - OK - warnings on some shadows funcs
I did not see the warnings?!
PASS 6, FAIL 0
17 (of 19) .m files have no tests.
> splines 1.2.7 - OK
PASS 75, FAIL 0
5 (of 12) .m files have no tests.
> statistics 1.2.4 - OK
PASS 219, FAIL 25 (this really looks bad)
52 (of 117) .m files have no tests.
> stk 2.2.0 - OK
PASS 576, FAIL 348 (many failed because of undefined variables)
82 (of 222) .m files have no tests.
> strings 1.1.0 - FAILED - pcre - Bug #44640
-
> struct 1.0.11 - OK
PASS 9, FAIL 0, XFAIL 1
0 (of 3) .m files have no tests.
> symbolic 2.1.0 - OK
Unable to test (popen2: process creation failed)
> tsa  4.2.9 - OK - warn on shadows a core lib function
Warnings during compile:
ls: /usr/local/: No such file or directory
ls: /bin/mexext: No such file or directory
PASS 17, FAIL 0
50 (of 55) .m files have no tests.
> vrml 1.0.13 - OK - warnings on continuation markers
I did not see the warnings?!
44 (of 44) .m files have no tests.
> windows 1.2.1 - FAILED - Array compile errors - Bug #44180
-


Summing up the current versions of Octave Forge packages I get the
following impression. Please don't take the categories too serious,
because I have no detailed knowledge of most packages. I just ran the tests.

4.0 COMPATIBLE
-bim
-communications
-ga
-geometry
-interval
-level-set (under the assumption that tests are deprecated)
-lssa
-msh (under the assumption that tests are deprecated)
-nan
-octcdf
-octclip
-optics
-splines
-struct

DEPRECATED
-java

NOT INSTALLABLE
-dicom
-fem-fenics
-fl-core
-image-acquisition
-ltfat
-mechanics
-mpi
-nurbs
-ocs
-octproj
-odepkg
-parallel
-secs2d
-strings
-windows

SERIOUS TEST FAILURES
-control
-dataframe
-financial
-image
-miscellaneous
-queueing
-signal
-statistics
-stk

UNKNOWN (INSUFFICIENT TESTS)
-cgi
-database
-data-smoothing
-divand
-econometrics
-fits
-fpl
-fuzzy-logic-toolkit
-general
-generate_html
-instrument-control
-io
-linear-algebra
-mvn
-nan
-ncarray
-netcdf
-optim
-optiminterp
-quaternion
-secs1d
-secs3d
-sockets
-specfun
-symbolic (I was unable to test)
-tsa
-vrml


Oliver

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Mike Miller-4
In reply to this post by JohnD
On Fri, Apr 10, 2015 at 16:05:15 -0400, JohnD wrote:
> I just tried installing all of packages that currently in the active
> list on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
> This is trying to install on the windows machine, not attempting cross
> compile, using what is available as the 'current' package on
> octave-forge.
> Heres the following results:

So do you intend to drop or patch the remaining failing packages in
build_packages.m (dicom, odepkg, windows) if no fixed releases are made
before the official 4.0 Windows binary is built?

Thanks for working on this,

--
mike

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Mike Miller-4
In reply to this post by Oliver Heimlich
On Sat, Apr 11, 2015 at 01:49:42 +0200, Oliver Heimlich wrote:
> Several warnings during compile: error: writing file 'C:\.octave_hist': Permission denied

This is bug #41131.

> Summing up the current versions of Octave Forge packages I get the following
> impression. Please don't take the categories too serious, because I have no
> detailed knowledge of most packages. I just ran the tests.

Would you mind sharing the test failures you are seeing? Either as bug
reports against each package or a summary log file sent to the list or
posted somewhere. Full output would be most helpful as many developers
are unable to test on Windows.

Thanks,

--
mike

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

JohnD
In reply to this post by Mike Miller-4
On 04/10/2015 09:11 PM, Mike Miller wrote:

> On Fri, Apr 10, 2015 at 16:05:15 -0400, JohnD wrote:
>> I just tried installing all of packages that currently in the active
>> list on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
>> This is trying to install on the windows machine, not attempting cross
>> compile, using what is available as the 'current' package on
>> octave-forge.
>> Heres the following results:
> So do you intend to drop or patch the remaining failing packages in
> build_packages.m (dicom, odepkg, windows) if no fixed releases are made
> before the official 4.0 Windows binary is built?
>
> Thanks for working on this,
>
Since there are patches available for most, I could patch most of them
in mxe.

The windows package may be a different story, as even patched, it
currently says it has errors.

But, even if patched in in mxe, anyone who tries installing from octave
forge will get failures.

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

tmacchant
In reply to this post by Oliver Heimlich




----- Original Message -----

> From: Oliver Heimlich
> To: octave-maintainers
> Cc:
> Date: 2015/4/11, Sat 08:49
> Subject: Re: [Octave Forge] Octave 4.0 call for packages
>
>
>
> On 10.04.2015 22:05, JohnD wrote:
>>
>>
>>>  -----Original Message-----
>>>  From: carandraug On Behalf Of
>>>  Carnë Draug
>>>  Sent: Tuesday, April 07, 2015 7:13 AM
>>>  To: John Donoghue
>>>  Cc: octave-maintainers; Mike Miller
>>>  Subject: Re: [Octave Forge] Octave 4.0 call for packages
>>>
>>>  On 6 April 2015 at 23:31, John Donoghue

>>>  wrote:
>>>>  On 04/06/2015 06:16 PM, [hidden email] wrote:
>>>>>
>>>>>  Message: 4
>>>>>  Date: Mon, 6 Apr 2015 21:17:38 +0100
>>>>>  From: Carn? Draug
>>>>>  To: Mike Miller>
>>>>>  Cc: Octave Maintainers
>>>>>  Subject: Re: [Octave Forge] Octave 4.0 call for packages
>>>>>  Message-ID:
>>>>>
>>>>>  <CAPOrs_1RNATmWYF9unM6NW7QDUxtJT24G-
>>>  [hidden email]>
>>>>>  Content-Type: text/plain; charset=UTF-8
>>>>>
>>>>>  On 15 March 2015 at 15:04, Mike
> Miller  wrote:
>>>>>>
>>>>>>>  [...]
>>>>>>>
>>>>>>>  The imminent Octave 4.0.0 release will be the first to
> supply an
>>>>>>>  official binary distribution of Octave for Windows
> operating systems.
>>>>>>>  This distribution will include a set of Octave Forge
> packages that
>>>>>>>  may be installed at the user's option. This is why
> we would like to
>>>>>>>  have compatible packages ready to incorporate into the
> next release
>>>>>>>  candidate for further testing.
>>>>>>>
>>>>>
>>>>>  Can users that have installed Octave with such installer then
> install
>>>>>  more packages and other versions of those packages?
>>>>>
>>>>>  If we are starting to distribute binaries of Octave core, would
> be
>>>>>  nice if we can also distribute binaries for the packages,
> specially
>>>>>  since the release cycle of core and OF is quite different.
>>>>>
>>>>>  Carn?
>>
>>  I just tried installing all of packages that currently in the active list
> on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
>>  This is trying to install on the windows machine, not attempting cross
> compile, using what is available as the 'current' package on
> octave-forge.
>>  Heres the following results:
>
> I can confirm your observations. I did an installation on Win 7 64bit
> and additionally did some “pkg test”ing. Unfortunately only few packages
> contain sufficient tests and allow an assessment. Here are my results
> (didn't check them against known bugs):
>
>>  bim  1.1.5      - OK
> PASS 92, FAIL 0
> 15 (of 47) .m files have no tests.
>>  cgi  0.1.0     - OK
> 6 (of 6) .m files have no tests.
>>  communications  1.2.1 -     OK - warns unusable help found
> PASS 457, FAIL 0
> 19 (of 115) .m files have no tests.
>>  control 2.8.0    - OK
> PASS 317, FAIL 13
> 229 (of 287) .m files have no tests.
>>  database  2.3.1     - OK
> PASS 0, FAIL 1 (should be expected)
> 5 (of 6) .m files have no tests.
>>  dataframe  1.1.0     - OK
> PASS 3, FAIL 9
> 101 (of 101) .m files have no tests.
>>  data-smoothing  1.3.0     - OK
> 4 (of 4) .m files have no tests.
>>  dicom      0.1.1    - FAILED - Error with octave_map - Bug #44624
> -
>>  divand  1.1.2    - OK
> PASS 1, FAIL 0
> 115 (of 116) .m files have no tests.
>>  econometrics 1.1.1     - OK
> 28 (of 28) .m files have no tests.
>>  fem-fenics  0.0.3  - FAILED - requires eigen3 lib in mxe-octave, so
> currently an expected failure
> -
>>  financial 0.4.0     - OK
> PASS 158, FAIL 9
> 23 (of 57) .m files have no tests.
>>  fits 1.0.5      - OK
> No m-files for testing.
>>  fl-core 1.0.0     - FAILED - has been patched in mxe-octave - Bug #41214
> -
>>  fpl  1.3.4     - OK
> PASS 0, FAIL 5 (failed calls to deprecated function files)
> 10 (of 10) .m files have no tests.
>>  fuzzy-logic-toolkit 0.4.4     - OK - warning about continuation markers
> depreciated
> I did not see a warning?!
> 50 (of 50) .m files have no tests.
>>  ga 0.10.0     - OK    - lots of warning about continuation markers
> depreciated
> I did not see a warning?!
> PASS 79, FAIL 0
> 16 (of 22) .m files have no tests.
>>  general  1.3.4     - OK
> PASS 25, FAIL 0, XFAIL 1
> 24 (of 28) .m files have no tests.
>>  generate_html 0.1.7    - OK
> 8 (of 8) .m files have no tests.
>>  geometry 1.7.0     - OK     - warning on printf args - warn on cant cleanup
> after self
> I can confirm the printf warnings only.
> PASS 282, FAIL 3 (missing python, should be expected)
> 188 (of 233) .m files have no tests.
>>  image   2.4.0     - OK
> PASS 631, FAIL 5, XFAIL 21
> 54 (of 137) .m files have no tests.
>>  image-acquisition 0.2.1    - FAILED -  requires v4l library no in
> mxe-octave so expected
> -
>>  instrument-control 0.2.1    - OK
> 27 (of 27) .m files have no tests
>>  interval  0.1.5    - OK
> PASS 7264, FAIL 0, XFAIL 18
> 0 (of 249) .m files have no tests.
>>  io 2.2.7        - OK
> PASS 11, FAIL 0
> 30 (of 32) .m files have no tests.
>>  java          - part of core now not tried
> -
>>  level-set  0.2.0    - OK
> Compile Warning: -fPIC ignored for target
> PASS 198, FAIL 6 ('ls_copy_sign' undefined errors)
> 2 (of 31) .m files have no tests.
>>  linear-algebra  2.2.2     -OK    - warnings on abstract class types
> Several warnings during compile:  writing file 'C:\.octave_hist':
> Permission denied
> PASS 19, FAIL 1 (function 'fsin' not found)
> 57 (of 69) .m files have no tests.
>>  lssa  0.1.2    - OK
> PASS 16, FAIL 0
> 0 (of 7) .m files have no tests.
>>  ltfat  2.0.1     - FAILED      - 'cc' not found - patched in mxe. -
> bug #43603
> -
>>  mechanics 1.3.1     - FAILED     - install failure  Bug #44804
> -
>>  miscellaneous 1.2.1    - OK
> PASS 384, FAIL 9 (strsplit errors)
> 16 (of 30) .m files have no tests.
>>  mpi 1.2.0     - FAILED     - mxe has no mpi lib, so expected
> -
>>  msh 1.0.10    - OK
> PASS 28, FAIL 1 (the gmesh subprocess exited abnormally)
> 9 (of 21) .m files have no tests.
>>  mvn 1.1.0    - OK
> PASS 1, FAIL 0
> 18 (of 19) .m files have no tests.
>>  nan 2.7.1     - OK    - warning -fPIC, shadow core funcs
> Warnings during Compile:
> ls: /usr/local/: No such file or directory
> ls: /bin/mexext: No such file or directory
> PASS 31, FAIL 0
> 65 (of 81) .m files have no tests.
>>  ncarray  1.0.3     - OK
> 45 (of 45) .m files have no tests.
>>  netcdf 1.0.6    - OK
> Several warnings during compile: error: writing file
> 'C:\.octave_hist':
> Permission denied
> 10 (of 10) .m files have no tests.
>>  nurbs 1.3.9     - FAILED     - matrix_value - bug #44651
> -
>>  ocs  0.1.3     - OK    - continuation marker warnings
> Can't install, depends on odepkg
>>  octcdf 1.1.8    - OK
> Several warnings during compile: error: writing file
> 'C:\.octave_hist':
> Permission denied
> PASS 23, FAIL 0
> 30 (of 31) .m files have no tests.
>>  octclip 1.0.3     - OK    - printf format warnings
> PASS 5, FAIL 0
> 0 (of 1) .m files have no tests.
>>  octproj 1.1.2     - FAILED     - mxe-octave needs libproj, so expected
> failure
> -
>>  odepkg  0.8.4     - FAILED    -  octave_map - bug #44594
> -
>>  optics 0.1.1    - DONE
> PASS 166, FAIL 2 (%!test should be %!error)
> 0 (of 43) .m files have no tests.
>>  optim 1.4.1    - DONE
> PASS 16, FAIL 0
> 49 (of 54) .m files have no tests.
>>  optiminterp 0.3.4    - DONE
> Several warnings during compile: error: writing file
> 'C:\.octave_hist':
> Permission denied
> 9 (of 9) .m files have no tests.
>>  parallel 2.2.1     - FAILED     - bug #41148
> -
>>  quaternion 2.4.0    - OK
> PASS 23, FAIL 0
> 60 (of 76) .m files have no tests.
>>  queueing 1.2.3    - OK
> PASS 261, FAIL 4 ('sumexpn' undefined, and low tolerances), XFAIL 3
> 23 (of 101) .m files have no tests.
>>  secs1d 0.0.9     - OK
> 5 (of 5) .m files have no tests.
>>  secs2d 0.0.8     - FAILED     - Octave_map error Bug #44803
> -
>>  secs3d 0.0.1    - OK
> 39 (of 39) .m files have no tests.
>>  signal 1.3.1    - OK
> PASS 469, FAIL 3
> 78 (of 146) .m files have no tests.
>>  sockets 1.2.0    - OK
> No m-files for testing
>>  specfun 1.1.0    - OK     - warnings on some shadows funcs
> I did not see the warnings?!
> PASS 6, FAIL 0
> 17 (of 19) .m files have no tests.
>>  splines 1.2.7    - OK
> PASS 75, FAIL 0
> 5 (of 12) .m files have no tests.
>>  statistics 1.2.4    - OK
> PASS 219, FAIL 25 (this really looks bad)
> 52 (of 117) .m files have no tests.
>>  stk 2.2.0    - OK
> PASS 576, FAIL 348 (many failed because of undefined variables)
> 82 (of 222) .m files have no tests.
>>  strings 1.1.0     - FAILED     - pcre - Bug #44640
> -
>>  struct 1.0.11    - OK
> PASS 9, FAIL 0, XFAIL 1
> 0 (of 3) .m files have no tests.
>>  symbolic 2.1.0    - OK
> Unable to test (popen2: process creation failed)
>>  tsa  4.2.9    - OK     - warn on shadows a core lib function
> Warnings during compile:
> ls: /usr/local/: No such file or directory
> ls: /bin/mexext: No such file or directory
> PASS 17, FAIL 0
> 50 (of 55) .m files have no tests.
>>  vrml 1.0.13     - OK    - warnings on continuation markers
> I did not see the warnings?!
> 44 (of 44) .m files have no tests.
>>  windows 1.2.1    - FAILED     - Array compile errors - Bug #44180
> -
>
>
> Summing up the current versions of Octave Forge packages I get the
> following impression. Please don't take the categories too serious,
> because I have no detailed knowledge of most packages. I just ran the tests.
>
> 4.0 COMPATIBLE
> -bim
> -communications
> -ga
> -geometry
> -interval
> -level-set (under the assumption that tests are deprecated)
> -lssa
> -msh (under the assumption that tests are deprecated)
> -nan
> -octcdf
> -octclip
> -optics
> -splines
> -struct
>
> DEPRECATED
> -java
>
> NOT INSTALLABLE
> -dicom
> -fem-fenics
> -fl-core
> -image-acquisition
> -ltfat
> -mechanics
> -mpi
> -nurbs
> -ocs
> -octproj
> -odepkg
> -parallel
> -secs2d
> -strings
> -windows
>
> SERIOUS TEST FAILURES
> -control
> -dataframe
> -financial
> -image
> -miscellaneous
> -queueing
> -signal
> -statistics
> -stk
>
> UNKNOWN (INSUFFICIENT TESTS)
> -cgi
> -database
> -data-smoothing
> -divand
> -econometrics
> -fits
> -fpl
> -fuzzy-logic-toolkit
> -general
> -generate_html
> -instrument-control
> -io
> -linear-algebra
> -mvn
> -nan
> -ncarray
> -netcdf
> -optim
> -optiminterp
> -quaternion
> -secs1d
> -secs3d
> -sockets
> -specfun
> -symbolic (I was unable to test)
> -tsa
> -vrml
>
>
> Oliver


Oliver


Would you mind to try patched version odepkg-0.8.4 for octave v4 on my website:
http://www.tatsuromatsuoka.com/octave/Eng/odepkg/


I have confirmed that can be installed on octave-4.0.0-rc2 binary.

Tatsuro


Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

jbect
In reply to this post by Oliver Heimlich
Le 11/04/2015 01:49, Oliver Heimlich a écrit :
>> stk 2.2.0    - OK
> PASS 576, FAIL 348 (many failed because of undefined variables)
> 82 (of 222) .m files have no tests.

Ouch, that is a very high number of failing tests :((

I have never seen any "undefined variables" failure before and the tests
results sent by JohnD for 4.0.0rc2 32 bits show no failure for STK.

So, this must be a Win7/64bits-specific problem, then...

I don't have access to a Win7/64bits box at home, but I will be able to
test next week (hopefully on 4.0.0rc3, if the installer has been
released by then).

@++
Julien


Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Oliver Heimlich
In reply to this post by Mike Miller-4
On 11.04.2015 03:25, Mike Miller wrote:
> On Sat, Apr 11, 2015 at 01:49:42 +0200, Oliver Heimlich wrote:
>> Several warnings during compile: error: writing file 'C:\.octave_hist': Permission denied
>
> This is bug #41131.

Thanks for the info.


>> Summing up the current versions of Octave Forge packages I get the following
>> impression. Please don't take the categories too serious, because I have no
>> detailed knowledge of most packages. I just ran the tests.
>
> Would you mind sharing the test failures you are seeing? Either as bug
> reports against each package or a summary log file sent to the list or
> posted somewhere. Full output would be most helpful as many developers
> are unable to test on Windows.
>
> Thanks,

I am going to put the results into our wiki [1] as a table. I can put
the test logs on a private website and link to them in the table. Then,
we can collect references to existing bug reports/patches in a
structured way. This should simplify the assessment of all packages
together. And it might help to sort out deprecated packages.


On 06.04.2015 22:17, Carnë Draug wrote:> On 15 March 2015 at 15:04, Mike
Miller <[hidden email]> wrote:
 >> [...]
 >>
 >> The imminent Octave 4.0.0 release will be the first to supply an
 >> official binary distribution of Octave for Windows operating systems.
 >> This distribution will include a set of Octave Forge packages that may
 >> be installed at the user's option. This is why we would like to have
 >> compatible packages ready to incorporate into the next release candidate
 >> for further testing.
 >>
 >
 > Can users that have installed Octave with such installer then install
 > more packages and other versions of those packages?
 >
 > If we are starting to distribute binaries of Octave core, would be nice
 > if we can also distribute binaries for the packages, specially since the
 > release cycle of core and OF is quite different.
 >
 > Carnë

This is an important point. The only benefit of pre-installed packages
IMHO is that the user can (1) be sure that the package is compatible and
(2) save some compilation time for package installation.

Contrariwise, we have the disadvantage that new users have to care less
about packages and maybe never find out that they can get more packages
or newer versions from Octave Forge.

Oliver

[1] probably going to be here: http://wiki.octave.org/Octave-Forge

Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Oliver Heimlich
On 11.04.2015 11:56, Oliver Heimlich wrote:

> On 11.04.2015 03:25, Mike Miller wrote:
>> Would you mind sharing the test failures you are seeing? Either as bug
>> reports against each package or a summary log file sent to the list or
>> posted somewhere. Full output would be most helpful as many developers
>> are unable to test on Windows.
>>
>> Thanks,
>
> I am going to put the results into our wiki [1] as a table. I can put
> the test logs on a private website and link to them in the table. Then,
> we can collect references to existing bug reports/patches in a
> structured way. This should simplify the assessment of all packages
> together. And it might help to sort out deprecated packages.

You can find the results in the wiki:
http://wiki.octave.org/Octave-Forge#GNU_Octave_4.0_compatibility_assessment

The wiki contains a link to my test logs from Win7.

So far, there are 13 packages that seem fit, 10 with bug reports or
known fixes. Many where I am unsure about their current state. Please
update the status of packages that you know better than I do.

Reply | Threaded
Open this post in threaded view
|

Windows package [WAS: [Octave Forge] Octave 4.0 call for packages]

Philip Nienhuis
In reply to this post by JohnD
John Donoghue-2 wrote
On 04/10/2015 09:11 PM, Mike Miller wrote:
> On Fri, Apr 10, 2015 at 16:05:15 -0400, JohnD wrote:
>> I just tried installing all of packages that currently in the active
>> list on octave forge in Windows using octave 4.0.0 rc2   (32 bit)
>> This is trying to install on the windows machine, not attempting cross
>> compile, using what is available as the 'current' package on
>> octave-forge.
>> Heres the following results:
> So do you intend to drop or patch the remaining failing packages in
> build_packages.m (dicom, odepkg, windows) if no fixed releases are made
> before the official 4.0 Windows binary is built?
>
> Thanks for working on this,
>
Since there are patches available for most, I could patch most of them
in mxe.

The windows package may be a different story, as even patched, it
currently says it has errors.

But, even if patched in in mxe, anyone who tries installing from octave
forge will get failures.
More info is in a related bug report #44180.
Short recap:
In Octave compiled for 32-bit, a slightly patched Windows package may give some COM errors (or maybe even just one) but otherwise seems to work fine.
That is, spreadsheet  I/O test scripts in the io package that invoke COM/ActiveX run w/o any issues.

For 64-bit Octave (--enable-windows-64 or--enable-64 configure options in MXE) the patched windows package will install fine but doesn't work and even crashes Octave hard.

Avinoam patched the package along the lines I sketched (to arrive at what I use myself); I think that patched windows package version could be included in the MXE installer as long as the latter is built without the --enable-windows-64 configure option.

Philip
Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Lukas Reichlin-4
In reply to this post by Oliver Heimlich
On 11.04.2015, at 14:07, Oliver Heimlich <[hidden email]> wrote:

> On 11.04.2015 11:56, Oliver Heimlich wrote:
>> On 11.04.2015 03:25, Mike Miller wrote:
>>> Would you mind sharing the test failures you are seeing? Either as bug
>>> reports against each package or a summary log file sent to the list or
>>> posted somewhere. Full output would be most helpful as many developers
>>> are unable to test on Windows.
>>>
>>> Thanks,
>>
>> I am going to put the results into our wiki [1] as a table. I can put
>> the test logs on a private website and link to them in the table. Then,
>> we can collect references to existing bug reports/patches in a
>> structured way. This should simplify the assessment of all packages
>> together. And it might help to sort out deprecated packages.
>
> You can find the results in the wiki:
> http://wiki.octave.org/Octave-Forge#GNU_Octave_4.0_compatibility_assessment
>
> The wiki contains a link to my test logs from Win7.
>
> So far, there are 13 packages that seem fit, 10 with bug reports or known fixes. Many where I am unsure about their current state. Please update the status of packages that you know better than I do.
>

Hi Oliver

Which BLAS flavour did you use for testing? I assume you used the OpenBLAS which is chosen by default in the Octave installer. Regarding the control package, it
is known [1] that non-reference BLAS versions can lead to failing tests.
Therefore, try to rerun the test, but this time select the Reference BLAS implementation upon Octave installation. Don't forget to reinstall the control package, too.

I've tested the rc2 binary on a 64bit Windows 7 machine. control-2.8.0 passed all the tests with Reference BLAS.

Lukas

[1]
http://octave.sourceforge.net/control/function/test_control.html


Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

jbect
In reply to this post by jbect
Le 11/04/2015 08:59, Julien Bect a écrit :

> Le 11/04/2015 01:49, Oliver Heimlich a écrit :
>>> stk 2.2.0    - OK
>> PASS 576, FAIL 348 (many failed because of undefined variables)
>> 82 (of 222) .m files have no tests.
>
> Ouch, that is a very high number of failing tests :((
>
> I have never seen any "undefined variables" failure before and the
> tests results sent by JohnD for 4.0.0rc2 32 bits show no failure for STK.
>
> So, this must be a Win7/64bits-specific problem, then...
>
> I don't have access to a Win7/64bits box at home, but I will be able
> to test next week (hopefully on 4.0.0rc3, if the installer has been
> released by then).

Thanks Oliver for uploading your logs.

Here is a typical example of the test failures you get with Octave
4.0rc2 under Win7/64bits:

***** shared x t u
  t = rand(3, 2);
  x = stk_dataframe(t);
!!!!! shared variable initialization failed

***** test u = apply(x, 1, @sum);
!!!!! test failed
'x' undefined near line 2 column 12

***** assert (isequal(u, sum(t, 1)))
!!!!! test failed
'u' undefined near line 2 column 17

...

Could you please try to see if there's anything wrong with the code in
my %!shared block ?

pkg load stk
t = rand (3, 2);
x = stk_dataframe (t);

Otherwise, does anybody have any idea about why this fails ? Is there
anything wrong with the way I use %!shared and %!test blocks ?

Finally, I can see on the wiki, in the last column : "Tests don't
cleanup temporary files". What does it mean ? How do you see that in the
log ?

@++
Julien



Reply | Threaded
Open this post in threaded view
|

Re: [Octave Forge] Octave 4.0 call for packages

Oliver Heimlich
On 11.04.2015 14:38, Julien Bect wrote:

> Le 11/04/2015 08:59, Julien Bect a écrit :
>> Le 11/04/2015 01:49, Oliver Heimlich a écrit :
>>>> stk 2.2.0    - OK
>>> PASS 576, FAIL 348 (many failed because of undefined variables)
>>> 82 (of 222) .m files have no tests.
>>
>> Ouch, that is a very high number of failing tests :((
>>
>> I have never seen any "undefined variables" failure before and the
>> tests results sent by JohnD for 4.0.0rc2 32 bits show no failure for STK.
>>
>> So, this must be a Win7/64bits-specific problem, then...
>>
>> I don't have access to a Win7/64bits box at home, but I will be able
>> to test next week (hopefully on 4.0.0rc3, if the installer has been
>> released by then).
>
> Thanks Oliver for uploading your logs.
>
> Here is a typical example of the test failures you get with Octave
> 4.0rc2 under Win7/64bits:
>
> ***** shared x t u
>   t = rand(3, 2);
>   x = stk_dataframe(t);
> !!!!! shared variable initialization failed
>
> ***** test u = apply(x, 1, @sum);
> !!!!! test failed
> 'x' undefined near line 2 column 12
>
> ***** assert (isequal(u, sum(t, 1)))
> !!!!! test failed
> 'u' undefined near line 2 column 17
>
> ...
>
> Could you please try to see if there's anything wrong with the code in
> my %!shared block ?
>
> pkg load stk
> t = rand (3, 2);
> x = stk_dataframe (t);

I can debug into this later. Currently I rerun all tests with rc3 and
the Reference BLAS library.

> Otherwise, does anybody have any idea about why this fails ? Is there
> anything wrong with the way I use %!shared and %!test blocks ?
>
> Finally, I can see on the wiki, in the last column : "Tests don't
> cleanup temporary files". What does it mean ? How do you see that in the
> log ?

The files showed up in my working directory after running the tests. I
don't think they are mentioned in the log. Maybe the files would be
cleaned up, if the tests didn't fail? I have to investigate further...

Oliver

1234