summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Ensured that enums cannot be declared with ...the_drow/ensured-that-enums-cannot-be-declared-wi-1410868467138Omer Katz2014-09-161-1/+2
|
* Merged in malthe/cffi/tmpdir-from-environ (pull request #46)Armin Rigo2014-09-030-0/+0
|\ | | | | | | Add support for alternative "tmpdir"
* | Mention 'CFFI_TMPDIR' in changelogtmpdir-from-environmalthe2014-09-021-2/+3
| |
* | Add support for 'CFFI_TMPDIR' environment variablemalthe2014-09-022-2/+5
| | | | | | | | | | | | | | If set, the value is used as the 'tmpdir' instead of the caller directory's '__pycache__'. This allows correct run-time operation when the package source is read-only.
* | The 'Verifier' class already falls back to '_caller_dir_pycache'malthe2014-09-021-2/+1
|/
* gcc 4.9 seems to like to inline this huge function. That's a bad ideaArmin Rigo2014-08-241-0/+10
| | | | in my honest opinion, and triggers warnings too.
* Merged in vbernat/cffi/fix/unaligned-access (pull request #45)Armin Rigo2014-08-240-0/+0
|\ | | | | | | Use memcpy() instead of unaligned accesses. The fix is relevant on some platforms which would complain in this case.
* | Fix precision loss warning when casting integer types.fix/unaligned-accessVincent Bernat2014-08-241-1/+1
| |
* | Use memcpy() instead of unaligned accesses.Vincent Bernat2014-08-231-56/+51
|/ | | | | | | | When a structure is packed, some members may be unaligned. Some architectures like Sparc are unable to handle those and will be terminated with a SIGBUS. This changeset use memcpy() to avoid any unaligned accesses.
* Merged in pjenvey/cffi (pull request #44)Armin Rigo2014-08-092-1/+16
|\ | | | | | | avoid deprecated imp.get_suffixes on >= 3.3
| * Merged in leethargo/cffi/parse_constant_plus (pull request #43)Armin Rigo2014-07-280-0/+0
| |\ | | | | | | | | | add parsing of constant with unary +
| * | add parsing of constant with unary +schwarz2014-07-272-1/+16
| |/
* | make this not publicPhilip Jenvey2014-07-281-3/+3
| |
* | avoid deprecated imp.get_suffixes on >= 3.3Philip Jenvey2014-07-271-6/+12
|/
* ARM64 support. Tests pass under qemu.Stefano Rivera2014-07-242-5/+7
| | | | Fixes #136. Thanks Will Newton <will.newton@linaro.org>.
* Oops. Using memcpy() here can be bogus because the addresses can overlap.Armin Rigo2014-07-061-1/+1
|
* hg merge release-0.8Armin Rigo2014-07-050-0/+0
|\
* | update MD5/SHAArmin Rigo2014-07-051-2/+2
| |
* | Prepare release 0.8.6Armin Rigo2014-07-056-9/+9
| |
* | hg merge defaultArmin Rigo2014-07-051-4/+4
|\ \ | |/
| * Fix for Windows, which doesn't have a reasonable snprintf()Armin Rigo2014-07-051-4/+4
| |
| * hg merge release-0.8Armin Rigo2014-07-050-0/+0
| |\ |/ /
| * hg merge release-0.8Armin Rigo2014-07-050-0/+0
| |\
* | | MD5/SHAArmin Rigo2014-07-051-2/+2
| | |
* | | Argh. The version 0.8.4 works with the backend "0.8.4" or "0.8", butArmin Rigo2014-07-058-21/+21
|/ / | | | | | | | | | | not "0.8.3". As a result, the release 0.8.4 is completely unusable. Get rid of the possibility to have some different version numbers in the backend, and prepare for 0.8.5...
* | Add the MD5/SHAArmin Rigo2014-07-051-2/+2
| |
* | Update to 0.8.4Armin Rigo2014-07-055-7/+8
| |
* | A workaround by Alex Gaynor for a bug in distutils that shows up on OS/XArmin Rigo2014-07-051-0/+2
| |
* | Update MD5/SHAArmin Rigo2014-07-051-2/+2
| |
* | hg merge defaultArmin Rigo2014-07-0518-210/+463
|\ \ | |/
| * prepare for 0.8.3Armin Rigo2014-07-056-7/+7
| |
| * Add malloc.h, needed for alloca() in this file.Armin Rigo2014-07-011-0/+1
| |
| * Attempt to copy ctypes' name-mangling capability "_name@num".Armin Rigo2014-06-301-1/+21
| |
| * Add a passing Windows testArmin Rigo2014-06-291-0/+15
| |
| * A test for checking the multithreaded safety of errno (and GetLastErrorArmin Rigo2014-06-251-0/+57
| | | | | | | | on Windows).
| * Avoid using ssize_t hereArmin Rigo2014-06-241-3/+3
| |
| * pom pom pomArmin Rigo2014-06-241-2/+2
| |
| * Generalize 105930aef686Armin Rigo2014-06-241-2/+3
| |
| * Skip "ssize_t" on Windows, where it is usually not defined anyway.Armin Rigo2014-06-241-0/+2
| |
| * Don't check MS_WIN32, it's not always defined even on MSVC.Armin Rigo2014-06-243-3/+3
| |
| * Add stdint.h as an explicit include on non-Windows too.Armin Rigo2014-06-243-0/+3
| |
| * Unify (manually) the three blocks of #ifs for Windows. Add a commentArmin Rigo2014-06-244-51/+67
| | | | | | | | to attempt to keep them from drifting apart in the future.
| * Transplant 08a4b8145f90Armin Rigo2014-06-241-1/+1
| |
| * Rephrase again a paragraphArmin Rigo2014-06-211-7/+9
| |
| * Merged in techtonik/cffi (pull request #40)Armin Rigo2014-06-211-15/+16
| |\ | | | | | | | | | Attempt to improve intro text for people not familiar with problem
| | * index.rst: typo, edited online with Bitbucketanatoly techtonik2014-06-161-1/+1
| | |
| | * docs: English improvement, thank to Daniel Holthanatoly techtonik2014-05-301-1/+1
| | |
| | * Attempt to improve intro text for people not familiar with problemanatoly techtonik2014-05-301-15/+16
| | |
| * | Ah, on http://bugs.python.org/issue21778 I got the answer as to whatArmin Rigo2014-06-161-6/+3
| | | | | | | | | | | | the expected way to call this is.
| * | Force the "strides" to be non-NULL. With a NULL strides, CPython >= 3.3Armin Rigo2014-06-161-2/+6
| | | | | | | | | | | | | | | seems to segfault when doing "mymemoryview[:5] = ffi.buffer(..)". I have no clue how we're supposed to use this messy interface.
View Lorry Controller Status