Re: [Plplot-devel] Status of our octave binding and examples for version 4.4.0 of octave

[Alan W. I. <Ala...@gm...>]

On 2018-08-13 08:41-0000 Arjen Markus wrote:

> Hi Alan,
>
>> However, I hope those comments (such as the possibility mentioned above of
>> building an octave MinGW-w64/MSYS2 package for yourself using the pacman
>> packaging information at
>> <https://github.com/Alexpux/MINGW-packages/tree/master/mingw-w64-octave-hg>)
>> are still of some use to you.
>>
> I have tried to build Octave under MinGW-w64/MSYS2 using that webpage. However:
> - I had to remove the references to the patches, as these do not seem to be available. No idea what the effect is
> - The build failed on a mismatch in the checksums:
>
> $ makepkg -sCLf
> ==> Making package: mingw-w64-octave-hg r19590.6d75f1683ce8-1 (Sun, Aug 12, 2018 15:56:07)
> ==> Checking runtime dependencies...
> ==> Checking buildtime dependencies...
> ==> Retrieving sources...
>  -> Updating gnulib git repo...
> Fetching origin
>  -> Found tip.tar.bz2
> ==> ERROR: Integrity checks (sha256) differ in size from the source array.
>
> This is the point where I stopped. I have no clue as to whether this is a serious issue or not.

Let's ignore octave for a bit and look at the more general picture.
If you like the goal (in its own right) of learning to build packages
for the MinGW-w64/MSYS2 platform, then I suggest you follow the
package build documentation examples under the heading "Re-building a
package" at <https://github.com/msys2/msys2/wiki/Creating-packages>.
Can you use those steps to build both their examples (flex and
python3)?  If not, then you should take the problem to the MSYS2
mailing list to get help with that documentation.  But if that general
procedure works for you, then the next obvious step is to help out
with the PLplot packaging at
<https://github.com/Alexpux/MINGW-packages/tree/master/mingw-w64-plplot>
by first building that package as is, fixing it according to your own
experience (e.g., by moving to Python 3, and by enabling qt), and
drawing Alex's attention to your improvements (likely via a pull
request) so they will be made part of the official Plplot package for
this platform.

Moving back to the octave case, I am pretty sure unless that
semi-official package is completely broken, that the same procedures
used to build flex, python3, and PLplot should also work fine for
octave without all the issues you encountered above.  Assuming you can
build that octave package, then the next steps would be to evaluate
the result for our needs, and assuming it does work for those needs
(or can easily be fixed), advocate making it an official package
that the PLplot package depends on.

> I guess an alternative could be to simply build Octave and install it without pacman, but I have not tried that yet.

I would advise against that since unofficial octave builds are not
going to help our octave users that much on this particular platform
and may turn out to be a black hole sucking up a lot of your time.
Instead, I think it would be better in the long-run to help get an
official octave package running for this platform as I have outlined
above.

Alan
__________________________
Alan W. Irwin

Programming affiliations with the FreeEOS equation-of-state
implementation for stellar interiors (freeeos.sf.net); the Time
Ephemerides project (timeephem.sf.net); PLplot scientific plotting
software package (plplot.sf.net); the libLASi project
(unifont.org/lasi); the Loads of Linux Links project (loll.sf.net);
and the Linux Brochure Project (lbproject.sf.net).
__________________________

Linux-powered Science
__________________________