[OSGeo-Discuss] GEOS 3.8.2 Released

[Alexandre Neto]

Thanks Paul,

In that case, there's no need or interest in keeping a separate build in
conda-forge, people can/should always go to the latest version.

All the best,

Alexandre Neto

A sábado, 24/04/2021, 01:10, Paul Ramsey <pramsey at cleverelephant.ca>
escreveu:

>
>
> On Sat, Apr 10, 2021 at 2:38 PM Alexandre Neto <senhor.neto at gmail.com>
> wrote:
>
>> Probably not the right place or topic to ask, but here it goes... Is
>> there a formal LTR version for GEOS?
>>
>
> No, we can barely handle the formality around cutting releases without
> getting all formal about "long term" versus other stuff. Mostly new
> releases are new features and faster old features, and only occasionally is
> there anything backwards changing (the 3.9 overlay had some minor behaviour
> changes, but not enough to warrant a 4.0 release).
>
>
>> Or the 3.8.2 patch is only happening because of all the brand new
>> precision feature on 3.9.x?
>>
>
> No, we try to backport things that are back-portable, because we recognize
> that distros sometimes peg versions, even though upgrading is usually a
> safe bet. 3.8.2 is just because that branch had accumulated a lot of fixes
> without a release.
>
> P
>
>
>>
>> The reason I ask is that conda-forge will skip this patch because there's
>> already an higher version available.
>>
>> I wouldn't mind creating and maintaining a LTR branch on GEOS feedstock
>> (the conda forge recipe), so that other packages (like QGIS) can use it if
>> necessary.
>>
>> Thanks,
>>
>> Alexandre Neto
>>
>>
>> A sábado, 10/04/2021, 19:25, Paul Ramsey <pramsey at cleverelephant.ca>
>> escreveu:
>>
>>> The latest stable release of the 3.8 series includes a number of fixes
>>> for crash cases, so worth immediate upgrade for anyone using the 3.8
>>> series.
>>>
>>> http://download.osgeo.org/geos/geos-3.8.2.tar.bz2
>>>
>>> Changes in 3.8.2
>>> 2021-04-10
>>>
>>> - Bug fixes / improvements
>>>
>>>   - Fix Voronoi polygons robustness issue (#976)
>>>   - Fix segfault in SimplePointInAreaLocator caused by casting
>>>     MultiPolygon to Polygon (#1047, Oliver Tan)
>>>   - DistanceOp against geometry with empty components
>>>     crashes (#1026, Paul Ramsey)
>>>   - Fix buffer fillet (JTS-526, #743, Paul Ramsey)
>>>   - Remove undefined behaviour in CAPI (#1021, Greg Troxel)
>>>   - WKT writing of MULTIPOINT with EMPTY component crash (#1027, Paul
>>> Ramsey)
>>>   - Fix buffering problem (#1022, JTS-525, Paul Ramsey)
>>>   - Fix segfault in GEOSInterpolate against empty eollections (#1055,
>>>     Sandro Santilli)
>>>   - Fix GEOSProjectNormalized return -1 on exception (#1058, Joris Van
>>> den Bossche)
>>>   - Fix memory management quirk in CAPI (#1050, Paul Ramsey)
>>>   - Allow build on Apple ARM64 (Taras Zakharko)
>>>   - Fix crash in GEOSPreparedIntersects and GEOSPreparedContains for
>>>     POINT EMPTY input (Even Rouault)
>>>
>>> Also note that the 3.9 series is compatible with 3.8 so if you have the
>>> opportunity to upgrade, you can take advantage of the many performance and
>>> reliability improvements in 3.9.
>>> _______________________________________________
>>> Discuss mailing list
>>> Discuss at lists.osgeo.org
>>> https://lists.osgeo.org/mailman/listinfo/discuss
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/discuss/attachments/20210424/245d3456/attachment.html>