8f6d75b | Karl Wette | 04 October 2020, 05:16:11 UTC | Fix license files so GitHub recognises dual license | 04 October 2020, 05:16:11 UTC |
f5b90ce | Karl Wette | 04 October 2020, 05:14:23 UTC | COPYING.MIT: fix whitespaces, reformat to 80-character lines | 04 October 2020, 05:14:23 UTC |
8761032 | Karl Wette | 04 October 2020, 04:14:45 UTC | Dual license this repository under the GPLv2 and MIT (Expat) licenses | 04 October 2020, 04:14:45 UTC |
73da476 | Karl Wette | 25 February 2020, 03:41:20 UTC | Add documentation and support sections | 25 February 2020, 03:41:20 UTC |
be5718e | Karl Wette | 17 February 2020, 03:31:51 UTC | Add README for SWIGLAL repository | 17 February 2020, 05:49:21 UTC |
2d3c1d0 | Duncan Macleod | 20 November 2019, 19:20:11 UTC | refactore generate_swig_iface.py to make it more modular, and improve self-documentation. This also moves the script into a top-level `/swig/` path, rather than being housed in the `/lal/swig/` directory | 23 November 2019, 10:46:29 UTC |
60cef88 | Karl Wette | 15 November 2019, 03:50:38 UTC | SWIGLALOmega::tagLIGOTimeGPS(const char *str): check for str != NULL | 15 November 2019, 04:49:42 UTC |
75afc60 | Kipp Cannon | 09 August 2019, 03:54:35 UTC | lal/src/date/Date.h: remove #include of LALStdLib - this *really* completes (for real, this time) the removal of the LALStatus structure from the date package - as is the habit, a number of other codes were relying on Date.h to provide the #include of LALStdlib, so a variety of things here and there need their #include's adjusted. - the changes are confined to lal itself. | 09 August 2019, 04:06:21 UTC |
1a937b3 | Karl Wette | 18 January 2019, 15:46:18 UTC | SWIG: declare all size_t dims/strides arrays with a explicit length - Declaring with length implicit from initialiser list may not be recognised by older compilers | 18 January 2019, 15:46:46 UTC |
4f01ad9 | Karl Wette | 08 October 2018, 05:28:46 UTC | SWIG: correctly implement and test transparent views/copies of LALStringVector | 08 October 2018, 15:34:45 UTC |
9d7d0bd | Karl Wette | 08 October 2018, 05:26:34 UTC | SWIG: add missing 'const' to 'strides' | 08 October 2018, 15:29:40 UTC |
2971365 | Karl Wette | 08 October 2018, 05:18:55 UTC | SWIG: no need to use $argnum for variables in 'typecheck' typemaps | 08 October 2018, 15:29:40 UTC |
1259f3e | Karl Wette | 04 October 2018, 09:15:10 UTC | SWIG: %swiglal_array_viewin...() must return dimensions of array for %swiglal_array_copyin...() | 08 October 2018, 15:29:40 UTC |
0a613cc | Karl Wette | 04 October 2018, 09:35:26 UTC | SWIG: allocate memory for array struct copy-in using %swiglal_new_array() | 08 October 2018, 10:23:41 UTC |
d6f5c17 | Karl Wette | 04 October 2018, 05:09:57 UTC | SWIG: should not need to Py_CLEAR(nparr) if PyArray_Converter() fails | 08 October 2018, 10:23:41 UTC |
ad01ded | Karl Wette | 08 October 2018, 10:19:20 UTC | SWIGTestLALPython.py: remove 'print(pickled)' statements to reduce output volume | 08 October 2018, 10:23:42 UTC |
264d0fe | Karl Wette | 08 October 2018, 10:22:08 UTC | SWIGTestLALPython.py: run through 2to3 | 08 October 2018, 10:23:41 UTC |
7163fe1 | Leo Singer | 29 August 2018, 16:31:55 UTC | Make some common LAL types picklable | 30 August 2018, 16:51:50 UTC |
eb89bbe | Karl Wette | 30 August 2017, 12:18:29 UTC | SWIGCommon: add SWIGLAL(OWNS_THIS_STRING(...)) macro | 31 August 2017, 08:23:08 UTC |
93f2e22 | Karl Wette | 12 September 2017, 09:44:50 UTC | SWIGCommon: define INT8_C() and UINT8_C() in order to parse integer constants | 12 September 2017, 09:44:50 UTC |
d38571c | Karl Wette | 12 September 2017, 08:53:45 UTC | SWIGCommon: add macro OUTPUT_OWNED_BY_1ST_ARG(...) for output-only arguments | 12 September 2017, 09:04:25 UTC |
c96b097 | Karl Wette | 12 September 2017, 08:53:05 UTC | SWIGCommon: rename macro OWNED_BY_1ST_ARG(...) to RETURN_OWNED_BY_1ST_ARG(...) | 12 September 2017, 09:04:25 UTC |
19f9bd4 | Karl Wette | 04 September 2017, 13:57:09 UTC | SWIGCommon.i: get const-casting right for const struct member assignment | 04 September 2017, 14:04:28 UTC |
cbe25f3 | Karl Wette | 03 September 2017, 09:54:51 UTC | SWIGPython.i: fix warnings when compiling against Python >= 3.0.0 - Ref #5735 | 03 September 2017, 16:44:58 UTC |
f136dec | Karl Wette | 03 September 2017, 09:26:04 UTC | lalsuite_swig.m4: require SWIG >= 3.0.9 if Python >= 3.0.0 - Revert changes to SWIGTestLALPython.py from commit bd0474c650edcd6b7d6ff5c974a95a41edea95f4 - Ref #5735 | 03 September 2017, 16:44:36 UTC |
fd3d4a6 | Karl Wette | 26 July 2017, 05:53:24 UTC | SWIG: resolve nested typedefs when handling destructor functions - This handles e.g. in LALPulsar XLALDestroySFTVector() destroying an SFTVector which is a typedef to COMPLEX8FrequencySeriesVector - In case of duplicate destructors, e.g. XLALDestroyREAL4Vector() and XLALDestroyREAL4Sequence() which both destroy a REAL4Vector, prefer the destructor containing the struct name | 26 July 2017, 05:53:24 UTC |
766c8ac | Karl Wette | 19 May 2017, 20:37:54 UTC | SWIGOctave: octave_base_value::print() became non-const in version 4.0.0 | 19 May 2017, 20:38:42 UTC |
5d90096 | Karl Wette | 14 March 2017, 18:20:18 UTC | SWIG: separate macro for array structs - Array structs being a struct which consists of just an array, as opposed to an struct which contains an array as well as other parameters - Use ARRAY_STRUCT_{1,2}D() macros in LALDatatypes.h for LAL array-like data structures; use ARRAY_{1,2}D() macros everywhere else | 14 March 2017, 18:30:23 UTC |
9bda82a | Karl Wette | 30 January 2017, 12:56:45 UTC | SWIG: support use of ccache-swig - Test whether SWIG works with CCACHE_CPP2=1, especially whether it can generate dependencies; otherwise disable with CCACHE_DISABLE=1 - Remove uses of #pragma SWIG, which SWIG manual (CCache.html) claims is a potential issue when using ccache-swig | 30 January 2017, 12:57:37 UTC |
1c2a292 | Karl Wette | 29 January 2017, 05:37:12 UTC | SWIGCommon.i: add some static casts to silence compiler warnings | 29 January 2017, 08:02:37 UTC |
48f8800 | Karl Wette | 23 January 2017, 05:14:21 UTC | SWIG: bump minimum required version to 2.0.12 - This version introduced the SWIG_OCTAVE_PREREQ() macro - Can now simplify Octave version checks in SWIGOctave.i | 26 January 2017, 03:53:49 UTC |
a86c971 | Karl Wette | 23 January 2017, 04:44:45 UTC | SWIG: add support for Octave 4.2 - Octave 4.2 requires SWIG 3.0.12 (not yet released) - Prefer command-line octave-cli to (potentially GUI) octave - Check output of `mkoctfile -p CXX` for any flags, e.g. -std=c++11 - Use Octave CXXFLAGS when checking headers, in case -std=* is present - Random.h: _RANDOM_H conflicts with /usr/include/c++/5/bits/random.h (!) - SWIGOctave.h: - toplev.h is already included by SWIG (and deprecated in Octave 4.2) - datevec.length() is deprecated, use datevec.numel() instead - Add as_double() and as_single() methods to array views | 23 January 2017, 06:49:26 UTC |
d522f31 | Leo Singer | 19 January 2017, 17:35:11 UTC | Move lal version check after lal import | 19 January 2017, 17:35:11 UTC |
6ef8780 | Karl Wette | 05 December 2016, 08:21:51 UTC | SWIGTestLALPython.py: replace 'isinstance(x, t)' with more strict 'type(x) is t' | 05 December 2016, 08:21:51 UTC |
421a910 | Kipp Cannon | 29 November 2016, 05:12:03 UTC | swig LIGOTimeGPS: replace __repr__() override - with implementation that satisfies eval(repr(x)) == x - https://docs.python.org/2/reference/datamodel.html#object.__repr__ - https://docs.python.org/2/library/functions.html#repr - refs #4908 | 01 December 2016, 04:34:29 UTC |
ed88032 | Kipp Cannon | 28 November 2016, 06:00:31 UTC | XLALGPSToStr(): strip trailing 0s and decimal point - has the potential to significantly reduce the size of log files and other ascii dumps of things - this also makes lal's swig-wrapped LIGOTimeGPS's __str__() compatible with the __str__() in glue's and pylal's versions - but mostly this is motivated by the discovery that porting the CacheEntry class from glue's LIGOTimeGPS to lal's has made cache file output wonky (the time and duration columns now have decimal points and lots of 0s in them) - why fix it this way instead of patching CacheEntry is given by the reasons above. - refs #4903 | 01 December 2016, 04:34:29 UTC |
f6ea208 | Karl Wette | 30 October 2016, 19:31:41 UTC | Revert "SWIG: call XLAL...VCSInfoCheck() version mismatch checking functions from module load" This reverts commit 0e77c74ee51d639f4f22e980192aeba4c0427cf6. | 30 October 2016, 20:14:26 UTC |
bcb585b | Karl Wette | 28 October 2016, 23:09:04 UTC | SWIG: allow passing of NULL string pointers - Using [] (Octave) or None (Python), as for double pointers | 30 October 2016, 12:48:27 UTC |
86ef45d | Karl Wette | 28 October 2016, 21:36:25 UTC | SWIG: test typemaps for strings and double pointers | 30 October 2016, 12:48:27 UTC |
98efa86 | Karl Wette | 28 October 2016, 23:09:46 UTC | SWIGOctave.i: minor comment edit | 30 October 2016, 12:48:27 UTC |
9d3074f | Karl Wette | 05 October 2016, 09:50:32 UTC | SWIG: move scripting-language-specific struct extensions into scripting-language-specific headers | 05 October 2016, 09:51:34 UTC |
b0b428b | Karl Wette | 05 October 2016, 09:49:00 UTC | SWIG: error on unrecognised scripting language | 05 October 2016, 09:51:34 UTC |
5180dfd | Karl Wette | 05 October 2016, 09:48:33 UTC | SWIG: minor documentation edits | 05 October 2016, 09:51:34 UTC |
0047ee4 | Karl Wette | 20 September 2016, 14:20:58 UTC | lal/swig/SWIGTestLALPython.py: simplified SWIG version check | 20 September 2016, 14:23:57 UTC |
0baa76d | Leo Singer | 15 July 2016, 02:55:32 UTC | Add unit test workaround for https://github.com/swig/swig/pull/617 There was a bug in swig that broke type coercion for overloaded division operators in Python 3. The bug was fixed in swig 3.0.9. However, our reference operating systems (Debian stable and SL7) still have swig 3.0.7. | 19 September 2016, 13:13:09 UTC |
89c5fa0 | Karl Wette | 14 September 2016, 15:01:00 UTC | lal/swig/generate_swig_iface.py: fix typo/bug | 14 September 2016, 15:01:27 UTC |
a38f75f | Karl Wette | 10 August 2016, 12:26:18 UTC | SWIG: enforce tagnames for typedef'd enums - Enums without tagnames can trigger bugs in SWIG, e.g. for an enum typedef enum { ... } myenum and a function void myfunction(const myenum *e); SWIG generates the incorrect cast (enum myenum const*) e whereas if the enum is given a tagname typedef enum tagmyenum { ... } myenum SWIG generates the correct cast (enum tagmyenum const*) e | 10 August 2016, 12:37:03 UTC |
9b48cf3 | Karl Wette | 09 August 2016, 09:50:38 UTC | SWIG: support Python's copy.copy() and copy.deepcopy() functions - Shallow copies are supported for non-opaque structs - Deep copies are supported for non-opaque structs without a destructor, which presumably can be trivially copied - Otherwise an XLAL_ENOSYS (not implemented) error is raised - Fixes #4546 | 09 August 2016, 09:50:38 UTC |
20003b9 | Karl Wette | 08 August 2016, 23:15:00 UTC | SWIG: add (shallow) copy constructors to all non-opaque structs - Remove no longer needed COPY_CONSTRUCTOR() macro | 09 August 2016, 09:47:02 UTC |
8999460 | Karl Wette | 08 August 2016, 22:48:07 UTC | SWIGCommon: rename %swiglal_struct_create_cdtors() to %swiglal_struct_extend() | 09 August 2016, 09:47:02 UTC |
c1ea650 | Karl Wette | 08 August 2016, 22:30:53 UTC | SWIGLALOmega: use XLALSetErrno() to silently signal an error to wrapper function | 09 August 2016, 09:43:49 UTC |
056f25c | Leo Singer | 26 July 2016, 19:18:22 UTC | Fix tautological comparison detected by gcc6 | 26 July 2016, 19:50:18 UTC |
cd9d747 | Karl Wette | 01 July 2016, 16:21:19 UTC | SWIG: fix return type of __hash__() function wrappers to a PyLong - This reverts commit 106a5f96f608219c9f14f8c1fc87f037b699d65b - Refs #4330 | 01 July 2016, 16:21:19 UTC |
2ff4cb8 | Leo Singer | 16 June 2016, 16:00:10 UTC | Port lal unit test suite to Python 3 | 16 June 2016, 16:00:37 UTC |
809f578 | Adam Mercer | 05 June 2016, 17:17:22 UTC | test is failing on OS X, comment out | 05 June 2016, 17:17:22 UTC |
b621ae2 | Karl Wette | 18 February 2016, 14:29:32 UTC | SWIG: call XLAL...VCSInfoCheck() version mismatch checking functions from module load | 18 February 2016, 14:29:32 UTC |
4f8c929 | Karl Wette | 17 February 2016, 18:55:26 UTC | SWIG: allow wrapping of fixed arrays of pointers | 18 February 2016, 14:22:34 UTC |
d7e498a | Karl Wette | 18 November 2015, 16:49:36 UTC | SWIG: explicitly cast array dimensions from size_t to array length type - reduces warnings with some compilers | 18 November 2015, 16:49:36 UTC |
cd04801 | Karl Wette | 18 November 2015, 15:46:04 UTC | SWIG: compatibility fixes for Octave 4.0.0 - Octave 4.0.0 requires SWIG 3.0.7 or later | 18 November 2015, 16:17:56 UTC |
15a5c40 | Karl Wette | 05 November 2015, 12:41:25 UTC | SWIG: fix a bug in commit bf04eae9fd91e4df3b6ac3ddafa462b22e627d49 | 05 November 2015, 12:41:25 UTC |
149ca98 | Karl Wette | 05 November 2015, 12:18:50 UTC | SWIG: extend LIGOTimeGPS typemaps to accept Python objects with .gpsSeconds and .gpsNanoSeconds attributes - With tests in SWIGTestLALPython.py - Refs #2721 | 05 November 2015, 12:23:14 UTC |
c7d1067 | Karl Wette | 05 November 2015, 07:49:57 UTC | SWIG: add typecheck typemaps to %swiglal_specialised_typemaps() - Note that typecheck take the *lowest* precedence, i.e. in overloaded functions, typechecks for all other types are tried first before these typechecks, and swiglal_specialised_TAGNAME() is the last typecheck attempted. This is important for e.g. LIGOTimeGPS, where for numerical precision the (LIGOTimeGPS / double) operator must take precedence over the (LIGOTimeGPS / LIGOTimeGPS) operator. | 05 November 2015, 12:23:13 UTC |
640acd1 | Karl Wette | 05 November 2015, 12:03:37 UTC | SWIG: run Octave/Python test scripts with unbuffered standard output/error | 05 November 2015, 12:23:13 UTC |
70a1a5b | Karl Wette | 30 October 2015, 12:38:12 UTC | SWIG: avoid "taking address of temporary" warnings on SL6 (#2) | 30 October 2015, 13:14:13 UTC |
0926c55 | Karl Wette | 29 October 2015, 17:20:46 UTC | SWIG: avoid "taking address of temporary" warnings on SL6 | 29 October 2015, 18:49:26 UTC |
b58fc59 | Karl Wette | 12 August 2015, 19:12:30 UTC | SWIG: add directive 'COPYINOUT_ARRAYS' | 12 August 2015, 19:28:28 UTC |
6456fc1 | Karl Wette | 12 August 2015, 13:15:43 UTC | SWIG: rename directive 'VIEWIN_STRUCTS' to 'VIEWIN_ARRAYS' | 12 August 2015, 19:14:06 UTC |
9bac430 | Karl Wette | 12 August 2015, 18:39:00 UTC | SWIGTestLALPython.py: remove unneeded semicolons | 12 August 2015, 19:14:06 UTC |
6e6afc0 | Karl Wette | 17 July 2015, 15:36:44 UTC | SWIG: rename some SWIG directives for improved clarity - RETURNS_PROPERTY() -> OWNED_BY_1ST_ARG() - ACQUIRES_OWNERSHIP() -> OWNS_THIS_ARG() | 20 July 2015, 15:59:06 UTC |
16f9714 | Karl Wette | 18 June 2015, 02:32:30 UTC | SWIG: fix return values for Python comparison operators - Python comparison operators must return Py_NotImplemented if they cannot compare two objects, rather than raising an error, as SWIG does by default - Use typemaps and a dummy argument to hack SWIG to return Py_NotImplemented in "cleanup" code, bypassing normal error return path - Refs #2215 | 19 June 2015, 00:11:17 UTC |
c4a327d | Karl Wette | 19 May 2015, 14:56:57 UTC | SWIG: use generate_swig_iface.py to apply typemaps ignoring 'int' returns - Otherwise cannot use RETURN_VALUE(int,...) to disable typemaps | 19 May 2015, 17:42:21 UTC |
081b4ed | Karl Wette | 18 April 2015, 23:07:11 UTC | SWIG: fix loss of precision in LIGOTimeGPS reverse multiplication operator - Special typemaps needed for __mul__() to handle (double * LIGOTimeGPS) - Also add high-precision LIGOTimeGPS tests to Octave test script - Refs #2036 | 18 April 2015, 23:07:11 UTC |
f3ce166 | Kipp Cannon | 15 April 2015, 21:43:38 UTC | XLALGPSMultiply(): rework to improve precision - still not returning exactly-rounded results, but better. disagreement with exactly-rounded results from bc <= 1 ns (off-by-one in least-significant digit) - note the swig bindings unit tests need to be updated in this patch | 16 April 2015, 18:16:59 UTC |
efc1c9e | Kipp Cannon | 14 April 2015, 17:18:30 UTC | lal swig LIGOTimeGPS: explain absence of in-place operators | 16 April 2015, 17:47:45 UTC |
96fe2dd | Kipp Cannon | 13 April 2015, 20:05:44 UTC | lal: fix loss of precision in LIGOTimeGPS swig bindings - the .__sub() and .__rsub__() operators were implemented using double precision floats as intermediate values, loosing precision. this patch removes the intermediate format. - the .__mul__() and .__div__() operators were using LIGOTimeGPS as intermediate values. this patch provides an alternate code path when the multiplying factor is a double precision float, avoiding the loss of precision when casting small doubles to LIGOTimeGPS. NOTE: reversed operations are still broken (can't figure out how to fix). - add unit tests to catch these problems in the future | 16 April 2015, 17:47:45 UTC |
583d9f2 | Karl Wette | 21 March 2015, 12:45:51 UTC | Revert "SWIG: add directives for wrapping REAL4VectorAligned in VectorMath.h" - This reverts commit 0abaa0abf5b97f686dc6d0e4a51e4b98e5b94451. - VectorMath functions are designed to be low-level, so calling them from SWIG is not required (and indeed not possible with the change to a lower-level API). - Refs #358 | 23 March 2015, 08:31:11 UTC |
af6a04f | Karl Wette | 19 March 2015, 13:44:28 UTC | SWIG: fix compatibility with Octave 3.2.4 - datenum() does not accept a single DATEVEC argument, so pass date vector to datenum() as individual arguments - datenum() returns more than one value, so only fail if no values are returned - Refs #1970 | 19 March 2015, 15:02:12 UTC |
435e404 | Karl Wette | 18 March 2015, 14:44:12 UTC | SWIG: do not use XLALFillBrokenDownTime() for Octave 'tm' struct conversions - For input maps, use Octave datevec -> Octave datenum -> time_t -> struct tm - For output maps, use struct tm -> Octave datevec as before, but only 6 elements - Update and streamline tests - Refs #1970 | 18 March 2015, 14:49:56 UTC |
dd5b586 | Karl Wette | 18 March 2015, 14:38:41 UTC | SWIG: cleanup and streamline 'tm' struct tests for Python | 18 March 2015, 14:49:46 UTC |
587fccc | Kipp Cannon | 17 March 2015, 17:14:33 UTC | lal swig: fix test of UTC <--> GPS conversion wrapper - 6-element input no longer accepted - output is now a tuple instead of a list - day-of-week and day-of-year are not necessarily repopulated internally so the tests now always provide correct values (they actually are, still, repopulated internally but that's a quirk of the implementation of the LAL time conversion routines and should not be relied upon by these tests) | 17 March 2015, 17:14:33 UTC |
94eb157 | Kipp Cannon | 11 March 2015, 15:00:23 UTC | lal swig: fix issues in handling of struct tm - do not disregard tm_wday and tm_yday values in input and do not rely XLALFillBrokenDownTime() to populate these values (probably not threadsafe) - disallow 6-element input, require Python time.struct_time-compatible input. - return struct tm value to Python as a tuple instead of a list (the time.struct_time type is immutable, calling code written to expect mutability will not interoperate with other Python libraries) | 16 March 2015, 19:41:35 UTC |
bbcca20 | Karl Wette | 04 March 2015, 12:25:04 UTC | SWIG: do not contaminate config.h with SWIG-specific defines - Use AC_CHECK_HEADER() for header checks, does not define HAVE_... symbols - Use AC_CHECK_DECL() for NumPy declaration checks, pass SWIGLAL_HAVE_... symbols on command line to Python wrapping build | 04 March 2015, 12:25:04 UTC |
c0bda41 | Karl Wette | 03 March 2015, 14:54:11 UTC | SWIG: add SWIGLAL(COPY_CONSTRUCTOR(TAGNAME)) general macro | 03 March 2015, 15:43:35 UTC |
874ab5a | Karl Wette | 03 March 2015, 12:22:12 UTC | SWIG: consistently use allocate and/or copy new instances and arrays macros - Do not need to pass struct typedef name to %swiglal_struct_create_cdtors() | 03 March 2015, 15:43:35 UTC |
a522e43 | Karl Wette | 03 March 2015, 12:18:44 UTC | SWIG: use XLALCalloc() in allocate and/or copy new instances and arrays macros | 03 March 2015, 15:43:35 UTC |
35e6bb5 | Karl Wette | 03 March 2015, 12:19:55 UTC | SWIG: reorder general macros: function-related first, then struct-related | 03 March 2015, 15:43:35 UTC |
fcb19b3 | Karl Wette | 27 February 2015, 15:28:43 UTC | SWIG: add directives for wrapping REAL4VectorAligned in VectorMath.h - with basic sanity tests for Octave and Python | 27 February 2015, 15:28:43 UTC |
de13ad3 | Karl Wette | 27 February 2015, 14:18:35 UTC | SWIG: add CAST_STRUCT_TO() directive | 27 February 2015, 15:24:35 UTC |
c55ae8a | Karl Wette | 27 February 2015, 14:17:25 UTC | SWIG: reformat interface file comments to 100-column lines | 27 February 2015, 15:24:35 UTC |
17deec0 | Karl Wette | 27 February 2015, 13:22:28 UTC | SWIG: move general SWIGLAL() macros to separate section at end of SWIGCommon.i | 27 February 2015, 15:24:35 UTC |
c0559d5 | Karl Wette | 23 January 2015, 23:06:04 UTC | SWIG: move test code into separate SWIGTest.c source files | 23 January 2015, 23:06:04 UTC |
fe2e82d | Karl Wette | 15 January 2015, 12:21:43 UTC | Integrate SWIG interface code into Doxygen documentation | 15 January 2015, 13:00:47 UTC |
6ebdd86 | Karl Wette | 02 December 2014, 16:50:44 UTC | SWIG: protect against dereferencing a null pointer $1 | 02 December 2014, 17:04:48 UTC |
0b1910e | Karl Wette | 21 November 2014, 16:20:17 UTC | SWIG: remove bundling of generated sources in distribution tarballs - does not work well enough with LALSuite packaging - sundry other minor changes to SWIG bindings | 22 November 2014, 23:26:19 UTC |
202f2ba | Karl Wette | 06 November 2014, 20:42:10 UTC | SWIG: declare Python swiglal_get_reference() function with SWIGINTERNINLINE - should suppress unused function warnings since SWIGINTERNINLINE includes SWIGUNUSED | 06 November 2014, 20:42:10 UTC |
1620ebf | Karl Wette | 16 October 2014, 16:56:38 UTC | SWIG: silence Octave warnings when compiling on SL7 | 21 October 2014, 10:23:47 UTC |
681040e | Karl Wette | 15 September 2014, 12:55:36 UTC | SWIG: fix bugs in getting pointers to Octave array data for viewing | 15 September 2014, 12:55:36 UTC |
a7684f5 | Karl Wette | 15 September 2014, 12:55:02 UTC | SWIG: use a typedef to check the size of structs when viewing arrays - swiglal_field_count_... #defines do not get passed to dependent SWIG modules, so typemaps will only work in library where array type is defined, e.g. REAL8Vector viewing will only work in LAL - Instead create a dummy typedef sizechk_t containing only fields that should be in array type, then check if it's the same size as NAME. Not completely foolproof because of alignment issues, but anyway... | 15 September 2014, 12:55:02 UTC |
b1a3e59 | Karl Wette | 02 September 2014, 17:11:27 UTC | SWIG: do not rename a symbol if re-name would start with a digit or _ - Symbol names starting with a digit are not allowed in Python, and names starting with a _ also do not seem to work with modules - Test by importing LAL_C_SI and LAL_180_PI constants | 02 September 2014, 17:13:35 UTC |
7378cd2 | Karl Wette | 01 September 2014, 12:20:44 UTC | SWIG: add macros for ignoring members of structs | 01 September 2014, 14:11:09 UTC |