aboutsummaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)Author
2019-09-24createOperations() sorting: sort operations that cannot be exported as PROJ ↵Even Rouault
pipelines at end
2019-09-23Document the autopromotion to 3D when mix of 2D/3D CRS in ↵Even Rouault
createOperations(), and test it (fixes #1623)
2019-09-22promote proj_assign_context to proj.h from proj_experimental.hsnowman2
2019-09-22Update to version 1.50 of the geodesic library.Charles Karney
* Allow arbitrarily complex polygons in geod_polygon_*. In the case of self-intersecting polygons the area is accumulated "algebraically", e.g., the areas of the 2 loops in a figure-8 polygon will partially cancel. * Simplify code by using C99 functions remainder and remquo. * More test coverage. Fixes to associated files: * src/pipeline.cpp invoke geod_init with f = es / (1 + sqrt(1 - es)) instead of (the less accurate) f = 1 - sqrt(1 - es) * src/apps/geod_set.cpp remove "#undef f" (a dangling relic?).
2019-09-19Require C99 compiler see #1621Charles Karney
cmake and autoconf now stipulate C99 change c89 to c99 in travis jobs remove HAVE_C99_MATH checks (unrelated) relax Visual Studio compatibility check in cmake/project-config-version.cmake.in (VS 2019 can use a VS 2015 library but not vice versa).
2019-09-19Merge pull request #1618 from cffk/c99-math-updateCharles Karney
Remove unneeded C99 compatibility functions from math.cpp and proj_math.h I'll do the clean up of the -std=c89 flags etc. as a separate pull request.
2019-09-18Delete proj_math.h. Replace includes by <math.h>; we'll see if anyoneCharles Karney
needs <limits.h>. Update scripts/reference_exported_symbols.txt and src/proj_symbol_rename.h.
2019-09-18math.cpp removed since its isnan isn't used. Keep proj_math.h (whichCharles Karney
just includes math.h and limits.h) since it's included in a score of places.
2019-09-18Get rid of dead code. The end result of this chain of commits is toCharles Karney
eliminate most of math.cpp. All that is left is the handling of isnan (and I've this because math.cpp notes that gie.c uses pj_isnan). geodesic.c now handles supplying C99 math functions internally and this can go away once C99 support is mandated.
2019-09-18Let geodesic.c use its own versions of C99 math functions if necessary.Charles Karney
2019-09-18Try removing C99 math functions from math.cpp (but leave isnan).Charles Karney
2019-09-17ell_set.cpp: avoid division by zero in R_lat_a case. Fixes ↵Even Rouault
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=16130
2019-09-17helmert: validate scale parameter to avoid later division by zero. Fixes ↵Even Rouault
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=16106
2019-09-17aea_e_inverse(): avoid calling asin() with invalid argument which results in ↵Even Rouault
NaN being propagated. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=15336
2019-09-17pj_Convert_Geocentric_To_Geodetic(): avoid division by zero on weird input ↵Even Rouault
coordinates and ellipsoid values. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=15148&
2019-09-17isea: avoid undefined behaviour on left shift of negative value. Fixes ↵Even Rouault
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=15009
2019-09-17pj_Convert_Geocentric_To_Geodetic(): avoid division by zero on weird input ↵Even Rouault
coordinates and ellipsoid values. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=14766
2019-09-17geos: avoid division by zero. Fixes ↵Even Rouault
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=14666
2019-09-17eqdc: avoid potential division by zero. Fixes ↵Even Rouault
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=17190
2019-09-17Fix floating-point number into integer warning in math.cppCharles Karney
2019-09-17Fix typo in geodesic.cCharles Karney
2019-09-17Add atanh, copysign, cbrt, remainder, remquo to math.cpp.Charles Karney
The supported C99 math functions provided by math.cpp are thus hypot log1p asinh atanh copysign cbrt remainder remquo round lround Make compiler checks in CMakeLists.txt and configure.ac consistent with this set. Make geodesic.c use the math.cpp defined (instead of the internally defined) versions of hypot atanh copysign cbrt This is keyed off the presence of the PROJ_LIB macro. I had at one time https://github.com/OSGeo/PROJ/pull/1425 suggested supplying an additional macro PROJ_COMPILATION when compiling geodesic.c. However, PROJ_LIB seems to fill the bill OK. The *next* version of geodesic.c (due out in a few weeks) will also use remainder remquo All of this is only of concern for C compilers without C99 support. So this may become an historical footnote at some point.
2019-09-16Merge pull request #1582 from rschmunk/patch-1Kristian Evers
Debug Bertin1953
2019-09-15Merge pull request #1614 from rouault/fixes_related_to_1597Kristian Evers
Fixes related to #1597
2019-09-15Merge pull request #1611 from rouault/fix1563Even Rouault
cs2cs: autopromote CRS to 3D when there's a mix of 2D and 3D (fixes #1563)
2019-09-15DatabaseContext::lookForGridInfo(): avoid setting PROJ context errno. Fixes ↵Even Rouault
issue with test/gigs/5208.gie with previous commit when ntf_r93.gsb grid is not available
2019-09-15createFromPROJString(): ignore +no_defs when instanciating a ↵Even Rouault
'+init=epsg:xxxx +no_defs' string (related to #1597)
2019-09-15Ingestion of +proj=somerc +type=crs: avoid adding twice alpha, gamma, lon_0 ↵Even Rouault
(related to #1597)
2019-09-15Merge pull request #1608 from ↵Kristian Evers
rouault/improve_transforms_fromto_wgs84_gXXXX_realizations Improvements in transformations from/to WGS 84 (Gxxxx) realizations and vertical <--> geog transormations
2019-09-13cs2cs: autopromote CRS to 3D when there's a mix of 2D and 3D (fixes #1563)Even Rouault
2019-09-13Fix support for +proj=ob_tran +o_proj=lonlat/latlong/latlon (only ↵Even Rouault
+o_proj=longlat worked) (fixes #1601)
2019-09-12createOperations(): when tranforming from a compoundCRS whose vertical ↵Even Rouault
component is a BoundCRS, do not apply the horizontal transformation twice
2019-09-12createOperations(): use more candidates when transforming between a ↵Even Rouault
geographic and vertical CRS For example when transforming from NAD83+NAVD88 height to WGS84, there is no transformation between NAVD88 height to WGS84. In that case, use all potential transformations from NAVD88 height to another geographic CRS for the vertical part.
2019-09-12Coordinate transformation: improve transformations from/to WGS84 (Gxxxx)Even Rouault
Currently very few transformations from/to WGS84 (Gxxxx) are registered in the EPSG database, and there isn't even transformations between WGS84 EPSG:4326 and those ones. Consequently transformations to those realizations often ended up as no-operation, whereas going through WGS84 EPSG:4326 will bring more meaningful results. So register those EPSG:4326<-->WGS 84 (Gxxx) null transformations, and when having WGS 84 (Gxxx) as source/target, consider EPSG:4326 as an intermediate. This change has no effect on the existing direct transformations from/to WGS 84 (Gxxx).
2019-09-12createOperations(): make sure sorting function is transitive (a < b and b < ↵Even Rouault
c --> a < c), to get consistent results
2019-09-12C API: add proj_crs_create_bound_vertical_crs_to_WGS84()Even Rouault
2019-09-12createOperations(): make it work when transforming between CompoundCRS and ↵Even Rouault
Geog3D when the DB has only VertCS to Geog2D This is needed to fix cases that would not work if using the promoteTo3D()/--3d functionnality just added per a6e1d72890615b42f54edad9b17acff8e7623844 In some cases, the EPSG database only contains a Geographic 2D CRS (like NAD83), without a 3D version. Consequently vertical transformations between that Geographic CRS and a Vertical CRS are only available with a 2D CRS code (kind of a bug in modelization by the way...). So when promoting the Geographic 2D CRS to a 3D one, we suddenly cannot find the available transformations any more. So in such situation, try to fallback to the 2D CRS to restore the capability to find the available transformations.
2019-09-12Merge pull request #1604 from rouault/fix_proj_trans_generic_unknown_timeEven Rouault
proj_trans_generic(): properly set coordinate time to HUGE_VAL when no value is passed to the function
2019-09-12proj_trans_generic(): properly set coordinate time to HUGE_VAL when no value ↵Even Rouault
is passed to the function
2019-09-11createOperations(): fix conversion from/to PROJ.4 CRS strings with ↵Even Rouault
non-ISO-cosher options and towgs84/nadgrids This actually fixes a regression introduced in PROJ 6.2.0 per 78302efb70eb4b49610cda6a60bf9ce39b82264f that made a conversion like EPSG:4326 to "+proj=something +towgs84/+nadgrids +over +type=crs" apply the towgs84/nadgrids operation twice.
2019-09-11API: add CRS::promoteTo3D(), proj_crs_promote_to_3D() and ↵Even Rouault
proj_crs_create_projected_3D_crs_from_2D() (fixes #1587) Also add a --3d switch to projinfo
2019-09-09Add copyright and license headerEven Rouault
2019-09-08Merge pull request #1585 from rouault/replace_2018_by_2019Kristian Evers
Use in API and utilities WKT2_2019 instead of WKT2_2018 (fixes #1578)
2019-09-08Use in API and utilities WKT2_2019 instead of WKT2_2018 (fixes #1518)Even Rouault
- C API: PJ_GUESSED_WKT2_2019 is added, PJ_GUESSED_WKT2_2018 aliased to it - C API: PJ_WKT2_2019[_SIMPLIFIED] is added, PJ_WKT2_2018[_SIMPLIFIED] alias to it - C++ API: similarly for WKTFormatter::Convention::WKT2_2019[_SIMPLIFIED] Those above changes should be fully backward API and ABI compatible. projinfo changes: - accept WKT2_2019 as value for -o switch. WKT2_2018 is still accepted (undocumented) - output now uses 'WKT2_2019 string:', so might break scripts that would rely on that. Other internal code references to WKT2_2018 changes to WKT2_2019, included in tests.
2019-09-06Merge pull request #1589 from rouault/fix_1574Even Rouault
PROJStringParser::createFromPROJString(): avoid potential infinite recursion (fixes #1574)
2019-09-05BUG: add SANITIZE_CTX for coordoperation is_instantiable and ↵snowman2
has_ballpark_transformation
2019-09-05PROJStringParser::createFromPROJString(): avoid potential infinite recursion ↵Even Rouault
(fixes #1574) The exact circumstances are a bit difficult to explain, but they involve using a non-default context, enabling proj_context_use_proj4_init_rules() on it, using proj_create(ctxt, "+init=epsg:XXXX +type=crs"), whereas PROJ_LIB is defined to a directory that has a 'epsg' file in it.
2019-09-05Merge pull request #1581 from jef-n/win-no-abi-libKristian Evers
stick with proj.lib on windows
2019-09-05Merge pull request #1584 from rouault/add_proj_create_ellipsoidal_3D_csEven Rouault
C API: add proj_create_ellipsoidal_3D_cs()
2019-09-05Merge pull request #1583 from rouault/constify_proj_create_crs_to_crs_from_pjEven Rouault
proj_create_crs_to_crs_from_pj(): make the PJ* arguments const PJ*
generated by cgit v1.2.3 (git 2.25.1) at 2026-01-05 14:42:50 +0000