16717
|
|
|
arigo |
16 years ago
|
|
|
16716
|
|
|
arigo |
16 years ago
|
|
|
16715
|
|
|
arigo |
16 years ago
|
|
|
16714
|
|
|
arigo |
16 years ago
|
|
|
16713
|
|
|
arigo |
16 years ago
|
|
|
16712
|
|
|
arigo |
16 years ago
|
|
|
16711
|
|
|
arigo |
16 years ago
|
|
|
16710
|
|
|
arigo |
16 years ago
|
|
|
16709
|
|
merging of svn+ssh://codespeak.net/svn/pypy/branch/pypy-more-rtti-inprogress/module/posix/test/test_posix2.py revisions 45507 to 46315:
------------------------------------------------------------------------ r46302 | arigo | 2007-09-04 13:04:55 +0200 (Tue, 04 Sep 2007) | 2 lines Skip on Windows. ------------------------------------------------------------------------ r46300 | arigo | 2007-09-04 12:47:37 +0200 (Tue, 04 Sep 2007) | 2 lines Skip tests for Windows. ------------------------------------------------------------------------ r46197 | fijal | 2007-08-30 16:05:49 +0200 (Thu, 30 Aug 2007) | 2 lines Damn. Make this check anything. Thanks xoraxax ------------------------------------------------------------------------ r46000 | pedronis | 2007-08-26 23:19:59 +0200 (Sun, 26 Aug 2007) | 3 lines hfs+ (default os x fs) has no sparse files ------------------------------------------------------------------------ r45986 | exarkun | 2007-08-25 17:05:20 +0200 (Sat, 25 Aug 2007) | 1 line implement os.times for posix ------------------------------------------------------------------------ r45710 | arigo | 2007-08-16 13:09:54 +0200 (Thu, 16 Aug 2007) | 2 lines Some more large file support: os.ftruncate(), os.lseek(). ------------------------------------------------------------------------ r45697 | arigo | 2007-08-16 09:15:31 +0200 (Thu, 16 Aug 2007) | 8 lines A more complete support for os.stat(). The RPython one now returns a Controlled object that supports both tuple-like indexing and the st_xxx attributes. The extra attributes are then exposed at app-level. Note that support for sub-second timestamps is there, but possibly a bit fragile. I disabled it because in CPython 2.4 by default we don't get sub-second timestamps out of os.stat() either. ------------------------------------------------------------------------ r45656 | arigo | 2007-08-14 18:10:44 +0200 (Tue, 14 Aug 2007) | 6 lines For backup purposes, in-progress: convert more of the os module to rtti and try to get rid of the rllib.ros module by making the native interfaces RPythonic. This looks quite good in my opinion - seems that we've finally learned a reasonable way to do external functions. ------------------------------------------------------------------------
|
arigo |
16 years ago
|
|
|
16708
|
|
merging of svn+ssh://codespeak.net/svn/pypy/branch/pypy-more-rtti-inprogress/rpython/module/ll_os.py revisions 45507 to 46315:
------------------------------------------------------------------------ r46290 | arigo | 2007-09-03 18:28:37 +0200 (Mon, 03 Sep 2007) | 3 lines Sprinkling underscores in front of C library function names on Windows, for ctypes to find the functions. ------------------------------------------------------------------------ r46286 | arigo | 2007-09-03 17:55:29 +0200 (Mon, 03 Sep 2007) | 2 lines Fix os.listdir() on Linux ------------------------------------------------------------------------ r46281 | arigo | 2007-09-03 17:07:15 +0200 (Mon, 03 Sep 2007) | 2 lines Fix os.utime() for Linux. ------------------------------------------------------------------------ r46280 | arigo | 2007-09-03 17:02:35 +0200 (Mon, 03 Sep 2007) | 4 lines os.utime(): * Windows support * sub-second resolution support (not for Windows) ------------------------------------------------------------------------ r46274 | arigo | 2007-09-03 14:27:46 +0200 (Mon, 03 Sep 2007) | 2 lines More progress, mostly test fixing. ------------------------------------------------------------------------ r46268 | arigo | 2007-09-03 13:40:26 +0200 (Mon, 03 Sep 2007) | 2 lines Windozify listdir() and pipe(). ------------------------------------------------------------------------ r46266 | arigo | 2007-09-03 12:38:56 +0200 (Mon, 03 Sep 2007) | 3 lines Make the RPython interface to os.mkdir() always take a mode argument, which we just ignore on Windows (CPython's os.mkdir() behaves like that too). ------------------------------------------------------------------------ r46247 | santagada | 2007-09-02 23:53:12 +0200 (Sun, 02 Sep 2007) | 2 lines in the wait manpage for os x and freebsd this is the right place for the signal header. ------------------------------------------------------------------------ r46244 | arigo | 2007-09-02 19:56:09 +0200 (Sun, 02 Sep 2007) | 2 lines test_ll_os.py passes on Windows. ------------------------------------------------------------------------ r46243 | arigo | 2007-09-02 19:50:02 +0200 (Sun, 02 Sep 2007) | 2 lines Start again with Windows porting. ------------------------------------------------------------------------ r46193 | fijal | 2007-08-30 14:39:49 +0200 (Thu, 30 Aug 2007) | 2 lines Fix translation ------------------------------------------------------------------------ r46188 | fijal | 2007-08-30 12:36:38 +0200 (Thu, 30 Aug 2007) | 2 lines Use more cool helpers ------------------------------------------------------------------------ r46185 | fijal | 2007-08-30 12:15:02 +0200 (Thu, 30 Aug 2007) | 2 lines Use cool helpers ------------------------------------------------------------------------ r46154 | arigo | 2007-08-29 13:45:11 +0200 (Wed, 29 Aug 2007) | 2 lines In-progress Windowsification. I hope I didn't break it on Linux. ------------------------------------------------------------------------ r46146 | arigo | 2007-08-29 13:13:22 +0200 (Wed, 29 Aug 2007) | 2 lines No os.getuid()/geteuid() on Windows. ------------------------------------------------------------------------ r45986 | exarkun | 2007-08-25 17:05:20 +0200 (Sat, 25 Aug 2007) | 1 line implement os.times for posix ------------------------------------------------------------------------ r45981 | fijal | 2007-08-25 14:32:54 +0200 (Sat, 25 Aug 2007) | 2 lines Fix name ------------------------------------------------------------------------ r45980 | pedronis | 2007-08-25 13:58:18 +0200 (Sat, 25 Aug 2007) | 3 lines support an idiom such that in the best case the original function needs to specified only in the @registering decorator ------------------------------------------------------------------------ r45979 | pedronis | 2007-08-25 13:19:16 +0200 (Sat, 25 Aug 2007) | 8 lines registering_if convenience decorator for the pattern: if hasattr(ns, name): @registering_if(ns.name) def ... ------------------------------------------------------------------------ r45976 | pedronis | 2007-08-25 12:51:46 +0200 (Sat, 25 Aug 2007) | 2 lines shorter names: lltypeimpl -> llimpl ------------------------------------------------------------------------ r45812 | arigo | 2007-08-17 15:05:56 +0200 (Fri, 17 Aug 2007) | 12 lines Change the level at which --sandbox'ing occurs. Now, by default the functions registered by _register_external() are replaced with a sandboxing version, so that the low-level external functions should no longer be seen at all. The _register_external()'ed functions operate at a slightly higher level - they have regular RPython objects as arguments and result. This is a more natural level that can be processed directly by pypy.rlib.rmarshal with no special-casing needed. This allows us to get rid of the strange marshal_input/unmarshal_output hacks in ll_os.py. As an extra bonus, the protocol used between the sandboxed process and its parent is now based on the regular CPython marshal format. ------------------------------------------------------------------------ r45735 | fijal | 2007-08-16 17:14:03 +0200 (Thu, 16 Aug 2007) | 3 lines Rename CXxx to CXxxPtr and make Cxx return xx (CStruct now returns a struct, not a pointer to this and so on) ------------------------------------------------------------------------ r45734 | arigo | 2007-08-16 16:56:52 +0200 (Thu, 16 Aug 2007) | 2 lines os._exit(). That was the last one, kill a couple of unnecessary files, yay! ------------------------------------------------------------------------ r45732 | arigo | 2007-08-16 16:46:47 +0200 (Thu, 16 Aug 2007) | 2 lines os.spawnv()... with a new test which I cannot run on Linux, so well, untested ------------------------------------------------------------------------ r45731 | arigo | 2007-08-16 16:34:59 +0200 (Thu, 16 Aug 2007) | 2 lines os.fork()... ------------------------------------------------------------------------ r45730 | arigo | 2007-08-16 16:32:52 +0200 (Thu, 16 Aug 2007) | 3 lines os.link()... os.symlink()... ------------------------------------------------------------------------ r45729 | arigo | 2007-08-16 16:27:20 +0200 (Thu, 16 Aug 2007) | 2 lines os.kill()... ------------------------------------------------------------------------ r45728 | arigo | 2007-08-16 16:21:08 +0200 (Thu, 16 Aug 2007) | 2 lines os.umask()... ------------------------------------------------------------------------ r45727 | arigo | 2007-08-16 16:19:34 +0200 (Thu, 16 Aug 2007) | 2 lines os.rename()... ------------------------------------------------------------------------ r45726 | arigo | 2007-08-16 16:17:57 +0200 (Thu, 16 Aug 2007) | 2 lines os.chmod()... ------------------------------------------------------------------------ r45725 | arigo | 2007-08-16 16:15:01 +0200 (Thu, 16 Aug 2007) | 4 lines os.chdir()... os.mkdir()... os.rmdir()... ------------------------------------------------------------------------ r45724 | arigo | 2007-08-16 16:06:42 +0200 (Thu, 16 Aug 2007) | 2 lines os.unlink()... ------------------------------------------------------------------------ r45723 | arigo | 2007-08-16 16:04:43 +0200 (Thu, 16 Aug 2007) | 2 lines os.system()... ------------------------------------------------------------------------ r45722 | arigo | 2007-08-16 15:59:07 +0200 (Thu, 16 Aug 2007) | 2 lines os.strerror()... ------------------------------------------------------------------------ r45721 | arigo | 2007-08-16 15:54:46 +0200 (Thu, 16 Aug 2007) | 2 lines os.isatty()... ------------------------------------------------------------------------ r45717 | arigo | 2007-08-16 14:08:13 +0200 (Thu, 16 Aug 2007) | 2 lines os.ftruncate(). Test large files when translated to C too. ------------------------------------------------------------------------ r45700 | arigo | 2007-08-16 11:40:30 +0200 (Thu, 16 Aug 2007) | 2 lines os.lseek -> rffi ------------------------------------------------------------------------ r45697 | arigo | 2007-08-16 09:15:31 +0200 (Thu, 16 Aug 2007) | 8 lines A more complete support for os.stat(). The RPython one now returns a Controlled object that supports both tuple-like indexing and the st_xxx attributes. The extra attributes are then exposed at app-level. Note that support for sub-second timestamps is there, but possibly a bit fragile. I disabled it because in CPython 2.4 by default we don't get sub-second timestamps out of os.stat() either. ------------------------------------------------------------------------ r45679 | arigo | 2007-08-15 15:03:44 +0200 (Wed, 15 Aug 2007) | 3 lines - support for 'os.environ.get(name)'. - kill rlib/ros.py. Yay! ------------------------------------------------------------------------ r45656 | arigo | 2007-08-14 18:10:44 +0200 (Tue, 14 Aug 2007) | 6 lines For backup purposes, in-progress: convert more of the os module to rtti and try to get rid of the rllib.ros module by making the native interfaces RPythonic. This looks quite good in my opinion - seems that we've finally learned a reasonable way to do external functions. ------------------------------------------------------------------------
|
arigo |
16 years ago
|
|
|
16707
|
|
|
arigo |
16 years ago
|
|
|
16706
|
|
|
arigo |
16 years ago
|
|
|
16705
|
|
|
arigo |
16 years ago
|
|
|
16704
|
|
|
arigo |
16 years ago
|
|
|
16703
|
|
|
arigo |
16 years ago
|
|
|
16702
|
|
|
arigo |
16 years ago
|
|
|
16701
|
|
merging of svn+ssh://codespeak.net/svn/pypy/branch/pypy-more-rtti-inprogress/module/posix/interp_posix.py revisions 45507 to 46315:
------------------------------------------------------------------------ r46019 | arigo | 2007-08-27 14:01:38 +0200 (Mon, 27 Aug 2007) | 3 lines In object spaces used during translation to OO backends, only expose the portable os.stat_result fields. ------------------------------------------------------------------------ r45986 | exarkun | 2007-08-25 17:05:20 +0200 (Sat, 25 Aug 2007) | 1 line implement os.times for posix ------------------------------------------------------------------------ r45828 | arigo | 2007-08-17 20:12:49 +0200 (Fri, 17 Aug 2007) | 2 lines don't capture the environment in the translated pypy ------------------------------------------------------------------------ r45710 | arigo | 2007-08-16 13:09:54 +0200 (Thu, 16 Aug 2007) | 2 lines Some more large file support: os.ftruncate(), os.lseek(). ------------------------------------------------------------------------ r45697 | arigo | 2007-08-16 09:15:31 +0200 (Thu, 16 Aug 2007) | 8 lines A more complete support for os.stat(). The RPython one now returns a Controlled object that supports both tuple-like indexing and the st_xxx attributes. The extra attributes are then exposed at app-level. Note that support for sub-second timestamps is there, but possibly a bit fragile. I disabled it because in CPython 2.4 by default we don't get sub-second timestamps out of os.stat() either. ------------------------------------------------------------------------ r45680 | arigo | 2007-08-15 15:35:51 +0200 (Wed, 15 Aug 2007) | 2 lines Missed the KeyError on del os.environ['unknown_name'] ------------------------------------------------------------------------ r45679 | arigo | 2007-08-15 15:03:44 +0200 (Wed, 15 Aug 2007) | 3 lines - support for 'os.environ.get(name)'. - kill rlib/ros.py. Yay! ------------------------------------------------------------------------ r45656 | arigo | 2007-08-14 18:10:44 +0200 (Tue, 14 Aug 2007) | 6 lines For backup purposes, in-progress: convert more of the os module to rtti and try to get rid of the rllib.ros module by making the native interfaces RPythonic. This looks quite good in my opinion - seems that we've finally learned a reasonable way to do external functions. ------------------------------------------------------------------------
|
arigo |
16 years ago
|
|
|
16700
|
|
|
arigo |
16 years ago
|
|
|
16699
|
|
merging of svn+ssh://codespeak.net/svn/pypy/branch/pypy-more-rtti-inprogress/config/translationoption.py revisions 45507 to 46315:
------------------------------------------------------------------------ r46032 | antocuni | 2007-08-27 16:08:52 +0200 (Mon, 27 Aug 2007) | 3 lines merge translator/{cli,jvm,oosupport} and config from dist ------------------------------------------------------------------------ r46028 | antocuni | 2007-08-27 15:51:13 +0200 (Mon, 27 Aug 2007) | 3 lines merge exceptiontransform from dist ------------------------------------------------------------------------ r46021 | antocuni | 2007-08-27 14:55:29 +0200 (Mon, 27 Aug 2007) | 3 lines merge dist to branch (only gencli-related changes) ------------------------------------------------------------------------ r45933 | arigo | 2007-08-23 15:41:45 +0200 (Thu, 23 Aug 2007) | 2 lines So, --thread forces gc=boehm for now. ------------------------------------------------------------------------ r45928 | arigo | 2007-08-23 14:25:34 +0200 (Thu, 23 Aug 2007) | 3 lines Try to clarify the memory allocation situation around thread.start_new_thread(). ------------------------------------------------------------------------ r45656 | arigo | 2007-08-14 18:10:44 +0200 (Tue, 14 Aug 2007) | 6 lines For backup purposes, in-progress: convert more of the os module to rtti and try to get rid of the rllib.ros module by making the native interfaces RPythonic. This looks quite good in my opinion - seems that we've finally learned a reasonable way to do external functions. ------------------------------------------------------------------------
|
arigo |
16 years ago
|
|
|
16698
|
|
|
arigo |
16 years ago
|
|
|