Re: GSoc 2016: Final Reviews required

[John Swensen]

> On Aug 20, 2016, at 4:00 AM, PhilipNienhuis <address@hidden> wrote:
> 
> AMR_KELEG wrote
>> 
>> PhilipNienhuis wrote
>>> 
>>> AMR_KELEG wrote
>>>> Dear all,
>>>> 
>>>> 
>>>> First of all, I would like to thank all of you for giving me this great
>>>> opportunity to contribute to the GNU Octave project.
>>>> I enjoyed my summer this year working on this project with all the ups
>>>> and downs - motives and frustrations.
>>>> 
>>>> Special thanks goes to my Mentor Professor John Swenson.
>>>> Working with you was such a privilege.
>>>> 
>>>> 
>>>> I have to send my work to Google this week as part of my final
>>>> evaluation so i would like to share my clone of the Geometry package
>>>> with you.
>>>> 
>>>> https://bitbucket.org/amr_keleg/octave-geometry/src/76898074c0f2bbf77cc38c6ff27609800c1baf2d?at=default
>>>> Kindly contact me if you have reviews/required edits.
>>> Amr,
>>> 
>>> Some test results.
>>> 
>>> I installed the header files belonging to boost-1_61-0.7z in
>>> /usr/include. 134 MB on disk, quite a bit IMO. (entire boost download
>>> would expand to 555 MB,still more excessive).
>>> 
>>> Next I've cloned your repo and made a geometry-3.0.0 package from it
>>> (incl. bootstrap)and built geometry-3.0.0 on both Linux and Windows.
>>> In both OS-es, pkg install mentions that "dissolve" isn't found and that
>>> boost needs an upgrade to >= 1.6 (while I have 1.61.0).  Do I need to
>>> install additional boost stuff alongside just the header files?
>>> 
>>> Then I tried the examples on the Mathworks site.
>>> (http://nl.mathworks.com/help/map/ref/polybool.html).
>>> A little worrying is that when trying the second example set, Octave
>>> crashes hard on the poly2fv call in the sixth block (subplot 2, 3, 6),
>>> both on Linux and Windows.
>>> Can you manage to fix that, please?
>> Dear Philip,
>> 
>> I don't think you need to install additional boost stuff alongside the
>> header files.
>> This means that the dissolve header file that is extracted from the Boost
>> Geometry 1.60 extensions isn't working with Boost Geometry 1.61 .
>> IMO,I can add the new version of dissolve header file that is part of the
>> Boost Geometry 1.61 extensions to the package and add some checks to
>> choose between the two header files.
>> So,what do you think?
> 
> That might be a good idea. But see below as regards boost versions.
> 
> 
>> I tried the Matworks second test and it's working smoothly.
>> So,I attached a screenshot of the test/output figures.
>> Screenshot_from_2016-08-20_05-32-50.png
>> <http://octave.1599824.n4.nabble.com/file/n4679337/Screenshot_from_2016-08-20_05-32-50.png>
>>   
> 
> To be sure I uninstalled boost-1.61.0 and downgraded to / installed
> boost-1.60.0
> Sure enough, the Matlab example went without a hitch then.
> 
> It not very reassuring that the polygon operations do not work with current
> boost but only with the previous version..... :-)
> I'd prefer that the polygon operations work with newest boost versions.
> 
> BTW 
> I find that installing geometry-3.0.0 (beta) from your github account takes
> a very long time to finish (some minutes) *after* the last compile and copy
> step. Would you by any chance know what is happening then? Just copying the
> binary modules into place shouldn't need that much time IMO. 
> But maybe it is just the polybool compile step that takes so long;
> installing packages is multi-threaded these days so it could be that the
> copy step is just hanging until polybool compilation is finalized.
> 
> Anyway, as I announced earlier this week I've made a start with comparing
> the various function versions (ispolycw etc.) to see which ones would be the
> best and how those could be incorporated in the geometry/mapping packages.
> 
> Philip
> 
> 
> 
> 
> --
> View this message in context: 
> http://octave.1599824.n4.nabble.com/GSoc-2016-Final-Reviews-required-tp4679299p4679342.html
> Sent from the Octave - Maintainers mailing list archive at Nabble.com.
> 

I think we still need to make another tweak to the makefile.in. The only 
feature of the polygons that is version dependent is the self-intersection 
correction (they call it the “dissolve” algorithm). Everything else should work 
as-is. The dissolve function is currently in the development repositories, but 
not in the releases. I was under the impression that the 1.60 dissolve worked 
for both 1.60 and 1.61, but I guess that isn’t the case. 

There should be two scenarios, though both should work with different levels of 
functionality:

1) all functions should work without self-intersection correction for 
Boost.Geometry > 1.57
2) all functions should work with self-intersection correction for 
Boost.Geometry > 1.60

I will take a quick look and help Amr figure out why it isn’t doing the 
conditional compile of the self-intersection correction.

John