This is needed because gdal abuses geos by using the C++ API,
contrarary to geos recommendations, and because geos upstream thinks
that the C++ ABI is intrinsicially unstable and therefore changes the
shlib version every release.
== Build (Unix) ==
* Support custom buildflags in GRASS plugin (5850)
* GRASS 7.0.0 support (5852)
* fix detection of ECW SDK 5.1 on Mac (5867, 5390)
* Add support for Poppler >= 0.31.0 (6118)
== Port ==
* /vsitar/: remove useless validation test that prevents from opening valid .tar files (5864)
== Algorithms ==
* gdal_grid: fix crash in optimized mode with GCC 4.4 on 64bit (5987)
* Fix failure in GDALSuggestedWarpOut2() when computing output image size in case the top-left and bottom-right corners tranform to the same point (5980)
* Rasterize: add check to avoid burning negative x values (5641)
== GDAL core ==
* GMLJP2 reader: add compatibility with OGC CRS URL as found in GMLJP2v2 (5940)
* GMLJP2: on reading, don't do axis inversation if there's an explicit axisName requesting easting, northing order (5960)
* GMLJP2: add missing rangeParameters element to validate against GMLJP2 schema (5707)
== Utilities ==
* gdalbuildvrt: fix potential crash when using -b switch (6095)
* gdalserver: fix compilation with recent GNU libc (such as in Arch Linux) (6073)
* gdalwarp: emit error message if file specified with -cutline cannot be opened (5921)
* gdalwarp: fix 1.11.2 regression when invoking several times gdalwarp with several input files on the same target file, and when the input files have a nodata setting: only the last input file was warped onto the target dataset (5909)
* ogr2ogr: fix crash with -clipdst when a reprojection fails before (5973)
* ogr_layer_algebra.py: for Update, Clip and Erase, only creates attribute of input layer by default
One mismatched digest found in geography/libmemphis02:
# package libmemphis02
recorded SHA1 (memphis-0.2.3.tar.gz) = dbc2f61e49b996dc9ca91df0de9a08eb7adbfa9b
calculated SHA1 (memphis-0.2.3.tar.gz) = 85993bce12c3616fcf6e7682a70b9605883edec2
No changes were made to the libmemphis02 distinfo file
Otherwise, existing SHA1 digests verified and found to be the same on
the machine holding the existing distfiles (morden). All existing
SHA1 digests retained for now as an audit trail.
This option fixes a pgsql version, and we're not really set up to deal
with that. It makes postgis only build for that one version. So far,
I'm not aware of a reason to have pgsql support in gdal. So, just
turn it off, and we can figure out a way to deal if there is actually
a need.
Discussed with Filip during postgis2 packaging, and I think suggested
to Brook and Adam.
gdal links against the geos C++ library, which the geos docs say not
to do. (The C++ library has a new shlib name every release, because
humans cannot understand if changes to C++ sources cause ABI changes.)
* Fix infinite GDALOpen recursion with some VRTs (4835)
* Avoid destroying existing overviews (.aux/.rrd) (4831)
* Support recognising NaN better in CPLStrtod() (4799)
* Fix windows declaration for CPLGetErrorHandlerUserData() (4755)
* Fix compilation with Gentoo modified zlib 1.2.6 (4723)
* Look for libgeotiff headers in /usr/include/libgeotiff too (4706)
* Improve warning handling in CPLClearRecodeStubWarningFlags() (4650)
* Fix Solaris compilation bug (4705)
* MorphFromESRI(): compare SPHEROID/PRIMEM parms, not names (4673)
* Avoid iffy casting of OGRWkbGeometryType to int (4847)
* Fix thread safety issue with CPLOpenShared()
