summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Close the branch about to be mergedpython3-portArmin Rigo2012-08-120-0/+0
|
* Keep the original usages of the macro PyString_AS_STRING as a macro,Armin Rigo2012-08-121-13/+14
| | | | at least on Python 2.
* Better error message.Armin Rigo2012-08-121-2/+2
|
* Python 2.x compatArmin Rigo2012-08-121-2/+4
|
* Fixes fixesArmin Rigo2012-08-124-7/+10
|
* FixArmin Rigo2012-08-121-5/+13
|
* hack hack hackArmin Rigo2012-08-126-66/+54
|
* Fix test_c to pass both on Python 2.6-2.7 and on Python 3.3.Armin Rigo2012-08-122-95/+125
| | | | | Will not care about Python 3.0-3.2 for the test suite because of the u'' syntax used everywhere, which was re-added in Python 3.3.
* hg merge defaultArmin Rigo2012-08-122-1/+16
|\
| * More regularily in casts to floats.Armin Rigo2012-08-122-1/+16
| |
* | Start to tweak test_c for 3.3.Armin Rigo2012-08-121-3/+9
| |
* | FixesArmin Rigo2012-08-121-12/+16
| |
* | Tests pass on Linux Python 2.6Armin Rigo2012-08-123-4/+6
| |
* | Fixes for 2.6 compatArmin Rigo2012-08-122-18/+22
| |
* | hg merge defaultArmin Rigo2012-08-121-0/+1
|\ \ | |/
| * Add a missing testArmin Rigo2012-08-121-0/+1
| |
* | hg merge default, and try to resolve all conflicts.Armin Rigo2012-08-1228-1429/+2929
|\ \ | |/ | | | | Not tested so far.
| * Don't try to call the missing function from the test: it just failsArmin Rigo2012-08-121-3/+3
| | | | | | | | in the hard way, and anyway we know it can only fail here.
| * intermediate checkinArmin Rigo2012-08-121-6/+14
| |
| * Test and fix (thanks nphg)Armin Rigo2012-08-122-2/+10
| |
| * Fix the documentation: the module file names are no longer basedArmin Rigo2012-08-121-3/+4
| | | | | | | | on the MD5 but on the CRC of the input strings.
| * Try to fix the tests on OS/X, mainly by skipping the ABI tests thatArmin Rigo2012-08-121-5/+7
| | | | | | | | | | depend on the 'stdout' and 'stderr' symbols being present (under this name) in the libc.
| * Add two tests, one passing and one (probably definitely) skipped.Armin Rigo2012-08-122-0/+29
| |
| * Bah. Can't put the skip() in setup_method(), because test_ffi_backendArmin Rigo2012-08-111-3/+7
| | | | | | | | subclasses this class.
| * This test runs twice (in two subclasses), so we need to avoid the fileArmin Rigo2012-08-111-2/+3
| | | | | | | | name collision for Windows.
| * Pom pom pomArmin Rigo2012-08-111-0/+11
| |
| * Bah, handle_t is some kind of keyword on Windows. Can't use it in theArmin Rigo2012-08-111-4/+4
| | | | | | | | test.
| * A #define for the use of snprintf below.Armin Rigo2012-08-111-0/+1
| |
| * Fix for test_vgen:test_nonstandard_integer_types on Windows.Armin Rigo2012-08-111-1/+11
| |
| * A test, passing on Linux, checking that all nonstandardArmin Rigo2012-08-111-0/+14
| | | | | | | | integer types are indeed defined.
| * Windows tests start to pass with this hack.Armin Rigo2012-08-111-0/+6
| |
| * Attempting to collect export_symbols correctly, as needed for WindowsArmin Rigo2012-08-113-4/+24
| |
| * Fix test for Windows.Armin Rigo2012-08-111-0/+3
| |
| * Fix: never send NULL to PyErr_Format().Armin Rigo2012-08-111-6/+9
| |
| * Finally found out the "right" way to implement ffi.gc(), in just aArmin Rigo2012-08-095-1/+55
| | | | | | | | few lines of Python code using weakrefs with callbacks.
| * Test for load_library(None).Armin Rigo2012-08-081-1/+11
| |
| * Allow weakrefs to any cdata object, prompted by issue 12.Armin Rigo2012-08-082-60/+55
| |
| * Add an example of using a WeakKeyDictionary to keep aliveArmin Rigo2012-08-081-1/+14
| | | | | | | | more objects as long as some root object is alive.
| * Tentative: allow load_library(None).Armin Rigo2012-08-071-2/+6
| |
| * Fix testsArmin Rigo2012-08-071-2/+4
| |
| * Fix the testsArmin Rigo2012-08-071-1/+2
| |
| * C99 has complex types as standard, at least "_Complex double" andArmin Rigo2012-08-061-0/+34
| | | | | | | | "_Complex float". Added tests but not implemented so far.
| * Revert 0d5efadab0ac and use a different approach: reading a fieldArmin Rigo2012-08-055-33/+72
| | | | | | | | | | | | of type 'foo[0]' returns not a <cdata 'foo[0]'>, but a <cdata 'foo *'>. This is a bit of a special-case, but more consistent with what we can do with the cdata.
| * Workaround: allow out-of-bound array indexes if the array is 'type[0]'.Armin Rigo2012-08-055-6/+43
| |
| * Add a skipped testArmin Rigo2012-08-051-0/+23
| |
| * Unimplemented featuresArmin Rigo2012-08-051-0/+18
| |
| * Test and fix.Armin Rigo2012-08-045-7/+75
| |
| * Be more precise: only 'struct *'.Armin Rigo2012-08-041-1/+1
| |
| * verify() doesn't fully work with the ctypes backend. Complain clearlyArmin Rigo2012-08-041-1/+6
| | | | | | | | in this situation.
| * Side-effect: the ctypes backend now supports verify().Armin Rigo2012-08-041-2/+1
| |
View Lorry Controller Status