| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
Central conic projection implemented as 'ccon'.
|
|
* Replace pj_ell_set with reimplementation supporting ellipsoid inheritance
* remove unreachable code from pj_ell_set.c
* Swap steps, so ellps args are read first, in accordance with historical behaviour
* Add ellipsoid tests to CI targets
* Reduce some optimistic tolerances
OS/X appears to have a slightly off float handling, resulting in differences at the nanometer level. Switching to 10 nm.
|
|
|
|
Let gie use init-file syntax
|
|
|
|
Switch pj_factors to use proj_geoc_lat
|
|
Repair double cct io-blunder
|
|
|
|
|
|
|
|
These were accidentally dropped during PR #669
|
|
* Introduce geodetic-geocentric conversions, as PJ_xxx style conversion step and as API entry points
* minor improvements and minor bug squashing
|
|
|
|
|
|
|
|
Give definition and def'n line no. in "Invalid operation" message
|
|
A year or so ago, @micahcochran put quite some effort into rationalizing the PI usage in PROJ.4, by ensuring that a useful number of M_PI related constants were defined in projects.h.
But apparently adjlon.c was left behind still using its own set of definitions - perhaps because it bends the values slightly, to avoid unwanted sign switching near the date line.
In this PR the "bending trick" is reimplemented using the rationalized M_PI constants only.
|
|
* Support numerical factors only
* Make sure h positive. Improve some comments
* Let pole overshoot check have effect even for geocentric latitudes
* Factor-typological constants, now all returning false, for backwards compatibility
|
|
|
|
* Removed remaining traces of the builtin selftest system. Moved all functionality to test/gie
* Updated Appveyor and Travis build scripts
* Another appveyor script update
|
|
* Poder dual autochecking implementation
* Debugging aid: Improvements in PJ_vgridshift.c and gie.c
* Most likely, the bugbeing tripped is in the gridshift code, so. uncomment suspicious lines in deformation.gie and merge this to support the debugging effort
|
|
Remove selftests from projection files starting with v,w
|
|
|
|
Remove selftests from projection files starting with r,s,t,u
|
|
Remove selftests from projection files starting with l, m
|
|
Remove selftests from projection files starting with n,o,p,q
|
|
|
|
|
|
Remove selftests from projection files starting with h, i, j, k
|
|
|
|
|
|
grid area. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3960. Credit to OSS Fuzz. master only
|
|
NULL. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3645. Credit to OSS Fuzz
|
|
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3573 , https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3643 and https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=3641 . Credit to OSS Fuzz
|
|
|
|
Use HAVE_C99_MATH instead of _WIN32_ and __ANSI__
|
|
HAVE_C99_MATH was recently added to the build system as a safer way to
determine if non-ansi math functions are available on the current
system. Previously different combinations of tests including _WIN32_ and
__ANSI__ have been in use, but cases where that strategy has failed is
known. Hence this change to a hopefully more robust check of math
function availability.
|
|
proj_create is now indifferent whether or not proj definition terms start with a '+' character. Also, improve gie to support testing this.
|
|
* Ensure C89 compliance and enable more warnings
According to the contributing guidelines, the library is developed
strictly in ANSI C 89. However, this is not enforced.
Additionally, we enable more warnings: apart from -Wall and -Wextra we
enable a warning that makes sure all enumeration values are covered in
a switch statement.
When compiling with Clang, we also turn on the warnings -Wc99-extensions
and -Wc11-extensions.
* Enable all warnings that are used on Travis
This makes it easier to notice problems before pushing.
|
|
|
|
* enter proj_angular_input and proj_angular_output, exit proj_angular_left and proj_angular_right
* remove unused variable 'unit'
* In gie: remove unused func 'torad_if_needed', and add static keyword where needed
* In gie: add some comments
|
|
|
|
Change the initialization type of the some operations
|
|
|
|
CONVERSIONS that are not PROJECTIONS
|
|
Use enumerations where appropriate
|
|
Enumerations have the following advantages over #define:
- they clearly connect a variable and the allowed constants,
- the meaning of code is not obfuscated by integer values,
- they are visible to the compiler, which can warn about (possibly)
incorrect usage.
There should be no functional change.
|
|
Pipeline initialisation improvements
|
