]> granicus.if.org Git - python/log
python
24 years agoUpped default memory size to 16MB
Jack Jansen [Wed, 7 Jun 2000 20:08:43 +0000 (20:08 +0000)]
Upped default memory size to 16MB

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:13:41 +0000 (09:13 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
Changed the API names for setting the default encoding.
These are now in line with the other hooks API names
(no underscores).

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:13:21 +0000 (09:13 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
Change the default encoding to 'ascii' (it was previously
defined as UTF-8).

Note: The implementation still uses UTF-8 to implement
the buffer protocol, so C APIs will still see UTF-8. This
is on purpose: rather than fixing the Unicode implementation,
the C APIs should be made Unicode aware.

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:12:54 +0000 (09:12 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
The locale module is turned on per default.

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:12:30 +0000 (09:12 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
Added some more codec aliases. Some of them are needed by the
new locale.py encoding support.

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:12:09 +0000 (09:12 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
Added support to set the default encoding of strings
at startup time to the values defined by the C locale.

The sys.setdefaultencoding() API is deleted after having
set up the encoding, so that user code cannot subsequentely
change the setting. This effectively means that only site.py
may alter the default setting.

24 years agoMarc-Andre Lemburg <mal@lemburg.com>:
Marc-André Lemburg [Wed, 7 Jun 2000 09:11:40 +0000 (09:11 +0000)]
Marc-Andre Lemburg <mal@lemburg.com>:
Added a new locale name aliasing engine which also supports
locale encodings, a feature which is used by the new default
encoding support in site.py.

24 years agoNew codec which always raises an exception when used. This
Marc-André Lemburg [Wed, 7 Jun 2000 09:04:05 +0000 (09:04 +0000)]
New codec which always raises an exception when used. This
codec can be used to effectively switch off string coercion
to Unicode.

24 years agoAdded preliminary documentation for the winreg module,
Fred Drake [Wed, 7 Jun 2000 04:07:48 +0000 (04:07 +0000)]
Added preliminary documentation for the winreg module,
by Mark Hammond <markh@activestate.com>.

Limited markup & consistency revisions by FLD.

24 years agoPatch from Rene Liebscher: this adds "--help-foo" options to list the
Greg Ward [Wed, 7 Jun 2000 03:00:06 +0000 (03:00 +0000)]
Patch from Rene Liebscher: this adds "--help-foo" options to list the
values that "--foo" can take for various commands: eg. what formats for
"sdist" and "bdist", what compilers for "build_ext" and "build_clib".

I have *not* reviewed this patch; I'm checking it in as-is because it also
fixes a paper-bag-over-head bug in bdist.py, and because I won't have
time to review it properly for several days: so someone else can
test it for me, instead!

24 years agoAlways look for the system config file in the Distutils module directory,
Greg Ward [Wed, 7 Jun 2000 02:29:03 +0000 (02:29 +0000)]
Always look for the system config file in the Distutils module directory,
and call it "distutils.cfg" instead of "pydistutils.cfg" (personal
config files are still ".pydistutils.cfg" or "pydistutils.cfg", though).

24 years agoPatch from Rene Liebscher:
Greg Ward [Wed, 7 Jun 2000 02:26:19 +0000 (02:26 +0000)]
Patch from Rene Liebscher:
Look for personal config file in /home/greg on Windows, too: users will have
to set /home/greg to use this, so it's not something that many people will
use.  But if python-dev comes up with the "right way" to divine a
home directory on Windows, we can use that to set /home/greg and poof! --
personal Distutils config files on Windows.

24 years agocorrect case in python16.dsp to allow build with VS6
Trent Mick [Tue, 6 Jun 2000 23:29:09 +0000 (23:29 +0000)]
correct case in python16.dsp to allow build with VS6

24 years agoPatch from Trent Mick:
Andrew M. Kuchling [Tue, 6 Jun 2000 20:52:17 +0000 (20:52 +0000)]
Patch from Trent Mick:
Fix a small bug in posixmodule.c where a char* is being
dereferenced where it should not be.

24 years agoFirst crack at the Distutils "config" command. Unlike other commands, this
Greg Ward [Tue, 6 Jun 2000 02:57:07 +0000 (02:57 +0000)]
First crack at the Distutils "config" command.  Unlike other commands, this
one doesn't *do* anything by default; it's just there as a conduit for data
(eg. include dirs, libraries) from the user to the "build" commands.
However, it provides a couple of Autoconf-ish methods ('try_compile()',
'try_link()', 'try_run()') that derived, per-distribution "config" commands
can use to poke around the target system and see what's available.

Initial experimenst with mxDateTime indicate that higher-level methods are
necessary: analogs of Autoconf's AC_CHECK_HEADER, AC_CHECK_LIB will be
needed too (and that's just to probe the C/C++ system: how to probe the
Python system is wide open, and someday we'll have to worry about probing a
Java system too).

24 years agoFix 'reinitialize_command()' so it resets the 'have_run' flag for the
Greg Ward [Tue, 6 Jun 2000 02:52:36 +0000 (02:52 +0000)]
Fix 'reinitialize_command()' so it resets the 'have_run' flag for the
command being reinitialized to false.

24 years agoSupport for multiple distribution formats in one run.
Greg Ward [Tue, 6 Jun 2000 02:51:38 +0000 (02:51 +0000)]
Support for multiple distribution formats in one run.

24 years ago'get_outputs()' now returns an empty list instead of None.
Greg Ward [Tue, 6 Jun 2000 02:18:13 +0000 (02:18 +0000)]
'get_outputs()' now returns an empty list instead of None.

24 years agoAdded lib-tk to the standard sys.path initializer.
Jack Jansen [Mon, 5 Jun 2000 08:35:47 +0000 (08:35 +0000)]
Added lib-tk to the standard sys.path initializer.

24 years agoRemoved assorted old ifdefs.
Jack Jansen [Sun, 4 Jun 2000 22:11:10 +0000 (22:11 +0000)]
Removed assorted old ifdefs.

24 years agoRemoved support for very old CodeWarrior runtimes.
Jack Jansen [Sun, 4 Jun 2000 22:10:25 +0000 (22:10 +0000)]
Removed support for very old CodeWarrior runtimes.

24 years agoRemoved SYMANTEC_CFM68K support.
Jack Jansen [Sun, 4 Jun 2000 22:07:37 +0000 (22:07 +0000)]
Removed SYMANTEC_CFM68K support.

24 years agoRemoved USE_STDWIN support.
Jack Jansen [Sun, 4 Jun 2000 22:02:02 +0000 (22:02 +0000)]
Removed USE_STDWIN support.

24 years agoRemoved THINK_C support.
Jack Jansen [Sun, 4 Jun 2000 21:56:05 +0000 (21:56 +0000)]
Removed THINK_C support.

24 years agoDefine USE_GUSI2, not USE_GUSI1. PythonStandalone still doesn't compile due to tk...
Jack Jansen [Sun, 4 Jun 2000 21:53:12 +0000 (21:53 +0000)]
Define USE_GUSI2, not USE_GUSI1. PythonStandalone still doesn't compile due to tk issues, though, but it isn't important right now.

24 years agoRemoved THINK_C support.
Jack Jansen [Sun, 4 Jun 2000 21:51:36 +0000 (21:51 +0000)]
Removed THINK_C support.
Added a CARBON indicator to the version string.

24 years agoFixed syntax error.
Greg Ward [Sun, 4 Jun 2000 15:30:35 +0000 (15:30 +0000)]
Fixed syntax error.
Half-fixed RPM 2 compatibility:added 'rpm_base' option, which must be set
  (to eg. /usr/src/redhat on a stock Red Hat system) if rpm2_mode is on.
  Still not quite working, though.

24 years agoUse 'ensure_string_list()' for 'formats' option, so that it can be
Greg Ward [Sun, 4 Jun 2000 15:12:51 +0000 (15:12 +0000)]
Use 'ensure_string_list()' for 'formats' option, so that it can be
spelled sensibly in a config file.

24 years agoPatch from Harry Henry Gebel:
Greg Ward [Sun, 4 Jun 2000 15:00:34 +0000 (15:00 +0000)]
Patch from Harry Henry Gebel:
Fills in question marks in help

Reads scripts in from files rather than strings

Adds RPM 2 compatibility mode (untested). Use of this mode requires that
--bdist-base be specified because bdist_rpm has no way of detecting where
RPM wants to find spec files and source files. An unmodified RedHat 5.0
system would require '--bdist-base=/usr/src/RedHat'. (You would also have
to be root.) If the rpmrc file has been modified to allow RPMs to be built
by normal users then --build-base would need to be changed accordingly.

Formats the changelog.

GPW: tweaked formatting, added some editorial comments.

24 years agoAdded the 'ensure_*' methods from bdist_rpm; refactored 'ensure_filename()'
Greg Ward [Sun, 4 Jun 2000 14:21:28 +0000 (14:21 +0000)]
Added the 'ensure_*' methods from bdist_rpm; refactored 'ensure_filename()'
and added 'ensure_dirname()'.

24 years agoRemoved the 'ensure_*' methods -- they're just too handy too keep in one
Greg Ward [Sun, 4 Jun 2000 14:20:57 +0000 (14:20 +0000)]
Removed the 'ensure_*' methods -- they're just too handy too keep in one
command class, so they're now in the Command base class.

24 years agoRenamed 'modules' option to 'py_modules', for consistency with Distribution
Greg Ward [Sun, 4 Jun 2000 13:42:52 +0000 (13:42 +0000)]
Renamed 'modules' option to 'py_modules', for consistency with Distribution
(and in order to generate a more sensible error message cleanly).

24 years agoUse PyArg_ParseTuple and specify the method names, following a suggestion
Andrew M. Kuchling [Sat, 3 Jun 2000 20:43:43 +0000 (20:43 +0000)]
Use PyArg_ParseTuple and specify the method names, following a suggestion
   from Greg Stein

24 years agoAdd missing PyArg_NoArgs() calls to methods that didn't take arguments
Andrew M. Kuchling [Sat, 3 Jun 2000 19:41:42 +0000 (19:41 +0000)]
Add missing PyArg_NoArgs() calls to methods that didn't take arguments
   (Pointed out by Moshe Zadka)

24 years agoLatex formatting fixes
Andrew M. Kuchling [Sat, 3 Jun 2000 03:06:58 +0000 (03:06 +0000)]
Latex formatting fixes

24 years agoPatches from Moshe, w/ AMK's revisions
Andrew M. Kuchling [Sat, 3 Jun 2000 02:52:40 +0000 (02:52 +0000)]
Patches from Moshe, w/ AMK's revisions
Wrote Unicode section

24 years agoPatch from Harry Henry Gebel: fixes a bit of code that slipped by my
Greg Ward [Sat, 3 Jun 2000 01:03:55 +0000 (01:03 +0000)]
Patch from Harry Henry Gebel: fixes a bit of code that slipped by my
overhaul last night.

24 years agoCatch DistutilSetupError from the Distribution constructor.
Greg Ward [Sat, 3 Jun 2000 01:02:06 +0000 (01:02 +0000)]
Catch DistutilSetupError from the Distribution constructor.

24 years agoAdded a bunch of new globals in '_init_mac()' -- these will be needed to
Greg Ward [Sat, 3 Jun 2000 00:44:30 +0000 (00:44 +0000)]
Added a bunch of new globals in '_init_mac()' -- these will be needed to
  support the forthcoming Cygwin/Mingw32 GCC-on-Windows patch.
Standardized CVS id line.

24 years agoFirst version of the project that produces a working Carbon binary. The binary has...
Jack Jansen [Fri, 2 Jun 2000 21:40:07 +0000 (21:40 +0000)]
First version of the project that produces a working Carbon binary. The binary has only been tested on MacOS9, so far. This project needs CW IDE 4.1b4 or later.

24 years agoRevived various of the compatability routines and made them Carbon-compliant. This...
Jack Jansen [Fri, 2 Jun 2000 21:38:19 +0000 (21:38 +0000)]
Revived various of the compatability routines and made them Carbon-compliant. This is needed because the initial carbon-python does not use GUSI.

24 years agoRemoved hfsrunning() macro. This means MacPython won't run on MacOS 5 (oops, System...
Jack Jansen [Fri, 2 Jun 2000 21:36:39 +0000 (21:36 +0000)]
Removed hfsrunning() macro. This means MacPython won't run on MacOS 5 (oops, System 5) or earlier anymore:-)

24 years agoMade the core toolbox modules carbon-compatible using the new greylist feature of...
Jack Jansen [Fri, 2 Jun 2000 21:35:07 +0000 (21:35 +0000)]
Made the core toolbox modules carbon-compatible using the new greylist feature of bgen: non-carbon methods are still included in non-carbon MacPython. The issue of backward compatibility of Python code is still open.
Macmodule and macosmodule have also been carbonified. Some functionality is still missing there.

24 years agoReplaced GetFInfo and SetFInfo calls with FspGetFInfo and FspSetFInfo calls, which...
Jack Jansen [Fri, 2 Jun 2000 21:29:59 +0000 (21:29 +0000)]
Replaced GetFInfo and SetFInfo calls with FspGetFInfo and FspSetFInfo calls, which are carbon-compatible.

24 years agoRemoved (within an #ifdef) Carbon-incompatabilities:
Jack Jansen [Fri, 2 Jun 2000 21:28:52 +0000 (21:28 +0000)]
Removed (within an #ifdef) Carbon-incompatabilities:
- Don't call all the toolbox init routines.
- No balloon help, so removed help code from the initial dialog.

24 years agoRemoved (within an #ifdef) Carbon-incompatabilities:
Jack Jansen [Fri, 2 Jun 2000 21:27:11 +0000 (21:27 +0000)]
Removed (within an #ifdef) Carbon-incompatabilities:
- Use TickCount() in stead of LMGetTicks()
- Don't use xxxGetFile event filter
- Don't call SystemTask or HandleSysWindowEvent (is this also automatically handled under os9??)
- Don't scan event queue for cmd-. (to be fixed at some point).

24 years agoFixed callback function arguments to be carbon-compatible.
Jack Jansen [Fri, 2 Jun 2000 21:23:09 +0000 (21:23 +0000)]
Fixed callback function arguments to be carbon-compatible.
(Temporarily?) removed call to ProcessHighLevelEvent until we find out how to do this (Carbon only).

24 years agoConfig file for PythonStandCarbon (small non-shared MacPython for Carbon).
Jack Jansen [Fri, 2 Jun 2000 21:19:16 +0000 (21:19 +0000)]
Config file for PythonStandCarbon (small non-shared MacPython for Carbon).

24 years agoMade compatible with MSL Carbon libraries.
Jack Jansen [Fri, 2 Jun 2000 21:18:09 +0000 (21:18 +0000)]
Made compatible with MSL Carbon libraries.

24 years agomake sure the case is correct (damned case-sensitive languages :-)
Jeremy Hylton [Fri, 2 Jun 2000 02:41:47 +0000 (02:41 +0000)]
make sure the case is correct (damned case-sensitive languages :-)

24 years agoBumped version number to 0.9pre (there will be a couple of code snapshots
Greg Ward [Fri, 2 Jun 2000 02:23:42 +0000 (02:23 +0000)]
Bumped version number to 0.9pre (there will be a couple of code snapshots
before the real release, but I want to make it clear that a major new
release is on the way).

24 years agoOnly print debugging output if DEBUG true.
Greg Ward [Fri, 2 Jun 2000 02:01:51 +0000 (02:01 +0000)]
Only print debugging output if DEBUG true.

24 years agoOnly print debugging output if DEBUG true (and deleted some of the more
Greg Ward [Fri, 2 Jun 2000 01:59:33 +0000 (01:59 +0000)]
Only print debugging output if DEBUG true (and deleted some of the more
extraneous debug prints).

24 years agoUse Distribution method 'dump_option_dicts()' for debugging output, and only
Greg Ward [Fri, 2 Jun 2000 01:55:36 +0000 (01:55 +0000)]
Use Distribution method 'dump_option_dicts()' for debugging output, and only
do so if DEBUG is true.

24 years agoDitched the obsolete '_get_package_data()' method and its
Greg Ward [Fri, 2 Jun 2000 01:52:04 +0000 (01:52 +0000)]
Ditched the obsolete '_get_package_data()' method and its
'_check_*()' helpers.

24 years agoFairly massive overhaul to support getting RPM inputs (extra meta-data,
Greg Ward [Fri, 2 Jun 2000 01:49:58 +0000 (01:49 +0000)]
Fairly massive overhaul to support getting RPM inputs (extra meta-data,
prep/build/etc. scripts, doc files, dependency info) from a config file
rather than the dedicated "package_info" file.  (The idea is that
developers will provide RPM-specific info in the "[bdist_rpm]" section of
setup.cfg, but of course it could also be supplied in the other config
files, on the command line, or in the setup script -- or any mix of the
above.)

Major changes:
  * added a boatload of options to 'user_options' and
    'initialize_options()': 'distribution_name', 'group', 'release', ...
  * added 'finalize_package_data()', which takes the place of
    '_get_package_data()' -- except it's called from 'finalize_options()',
    not 'run()', so we have everything figured out before we actually run
    the command
  * added 'ensure_string()', 'ensure_string_list()', 'ensure_filename()';
    these take the place of '_check_string()' and friends.  (These actually
    look like really useful type-checking methods that could come in handy
    all over the Distutils; should consider moving them up to Command and
    using them in other command classes' 'finalize_options()' method for
    error-checking).
  * various cleanup, commentary, and adaptation to the new way of
    storing RPM info in '_make_spec_file()'

24 years agoReformatted and updated many docstrings.
Greg Ward [Fri, 2 Jun 2000 00:44:53 +0000 (00:44 +0000)]
Reformatted and updated many docstrings.

24 years agoTrent Mick <trentm@ActiveState.com>:
Fred Drake [Thu, 1 Jun 2000 18:37:36 +0000 (18:37 +0000)]
Trent Mick <trentm@ActiveState.com>:
This patch correct bounds checking in PyLong_FromLongLong. Currently, it does
not check properly for negative values when checking to see if the incoming
value fits in a long or unsigned long. This results in possible silent
truncation of the value for very large negative values.

24 years agoTrent Mick <trentm@activestate.com>:
Fred Drake [Thu, 1 Jun 2000 17:59:17 +0000 (17:59 +0000)]
Trent Mick <trentm@activestate.com>:
Fix test of the "math" module so it does not break on platforms that do
not offer rint(); just skip that portion of the test in that case.

24 years agoFredrik Lundh: here's the 96.6% version of SRE
Jeremy Hylton [Thu, 1 Jun 2000 17:39:12 +0000 (17:39 +0000)]
Fredrik Lundh: here's the 96.6% version of SRE

24 years agoadd new Python/exceptions.c to python16 project (courtesy Trent Mick)
Jeremy Hylton [Thu, 1 Jun 2000 15:47:08 +0000 (15:47 +0000)]
add new Python/exceptions.c to python16 project (courtesy Trent Mick)

24 years agoImprove TypeError exception message for list catenation.
Fred Drake [Thu, 1 Jun 2000 14:31:03 +0000 (14:31 +0000)]
Improve TypeError exception message for list catenation.

24 years agoMichael Hudson <mwh21@cam.ac.uk>:
Fred Drake [Thu, 1 Jun 2000 03:12:13 +0000 (03:12 +0000)]
Michael Hudson <mwh21@cam.ac.uk>:
Removed PyErr_BadArgument() calls and replaced them with more useful
error messages.

24 years agoFix bug reported by atkins@gweep.net; re.compile(r"[\100-\410]")
Andrew M. Kuchling [Thu, 1 Jun 2000 03:02:48 +0000 (03:02 +0000)]
Fix bug reported by atkins@gweep.net; re.compile(r"[\100-\410]")
   dumps core.  Solution: fix check_escape() to match its comment and
   use only the low 8 bits of the octal number.

24 years agoMichael Hudson <mwh21@cam.ac.uk>:
Fred Drake [Thu, 1 Jun 2000 02:02:46 +0000 (02:02 +0000)]
Michael Hudson <mwh21@cam.ac.uk>:
Removed PyErr_BadArgument() calls and replaced them with more useful
error messages.

24 years agopatch from Charles Waldman--
Jeremy Hylton [Thu, 1 Jun 2000 01:17:17 +0000 (01:17 +0000)]
patch from Charles Waldman--
define ThreadError (== thread.error); docs should be updated, too

24 years agoRemember the list of archive files created in 'make_distribution()'.
Greg Ward [Thu, 1 Jun 2000 01:10:56 +0000 (01:10 +0000)]
Remember the list of archive files created in 'make_distribution()'.
Added 'get_archive_files()' so outsiders can get their hands on that list.

24 years agoOops, 'reinitialize_command()' forgot to return the command object if didn't
Greg Ward [Thu, 1 Jun 2000 01:09:47 +0000 (01:09 +0000)]
Oops, 'reinitialize_command()' forgot to return the command object if didn't
need to be reinitialized -- fixed.

24 years agoAdded 'reinitialize_command()' method -- delegated to Distribution instance.
Greg Ward [Thu, 1 Jun 2000 01:08:52 +0000 (01:08 +0000)]
Added 'reinitialize_command()' method -- delegated to Distribution instance.
Ensure 'make_archive()' method returns archive filename.

24 years agoEnsure that 'make_archive()' returns the name of the new archive file.
Greg Ward [Thu, 1 Jun 2000 01:07:55 +0000 (01:07 +0000)]
Ensure that 'make_archive()' returns the name of the new archive file.

24 years agoMore tweaking to make this command act like other Distutils commands:
Greg Ward [Thu, 1 Jun 2000 00:40:25 +0000 (00:40 +0000)]
More tweaking to make this command act like other Distutils commands:
  * added "--bdist-base" option to parameterize where we build
    the RPM (comes from "bdist" by default: "build/bdist.<plat>")
  * simplified/cleaned up some code in 'run()' in the process of
    removing (most) hard-coded directory names
  * if "--spec-only", drop spec file in "dist" rather than "redhat"
    (directory name still hard-coded, though)
  * use 'reinitialize_command()' to fetch the "sdist" object to
    tweak before running "sdist" command
  * use 'self.copy_file()' method rather than 'copy_file()' function
  * cosmetic tweaks to comments, error messages

24 years agoRegularize options a bit:
Greg Ward [Wed, 31 May 2000 23:56:45 +0000 (23:56 +0000)]
Regularize options a bit:
  * help strings start with lowercase
  * added affirmative version of '--no-clean' and '--no-rpm-opt-flags',
    which are the default (thus the attributes that correspond to
    the options are now 'clean' and 'use_rpm_opt_flags')

24 years agoDo not expose __builtins__ name as a completion; this is an implementation
Fred Drake [Wed, 31 May 2000 14:31:00 +0000 (14:31 +0000)]
Do not expose __builtins__ name as a completion; this is an implementation
detail that confuses too many people.  Based on discussion in python-dev.

24 years agoFilled out the "Core Changes" section.
Andrew M. Kuchling [Wed, 31 May 2000 03:28:42 +0000 (03:28 +0000)]
Filled out the "Core Changes" section.

24 years agoRenamed 'native_path()' to 'convert_path()'.
Greg Ward [Wed, 31 May 2000 02:32:10 +0000 (02:32 +0000)]
Renamed 'native_path()' to 'convert_path()'.
Also changed it so it doesn't barf if the path is already in native format
  (ie. contains os.sep).

24 years agoNormalize paths before writing them to a zipfile.
Greg Ward [Wed, 31 May 2000 02:17:19 +0000 (02:17 +0000)]
Normalize paths before writing them to a zipfile.

24 years agoFixed 'change_root() to work at all on Windows, and to work correctly on Unix.
Greg Ward [Wed, 31 May 2000 02:14:32 +0000 (02:14 +0000)]
Fixed 'change_root() to work at all on Windows, and to work correctly on Unix.

24 years agoImport the new Extension class, so setup scripts can
Greg Ward [Wed, 31 May 2000 01:11:20 +0000 (01:11 +0000)]
Import the new Extension class, so setup scripts can
"from distutils.core import" it.

24 years agoOverhauled to expect 'self.extensions' (taken from 'ext_modules' in the
Greg Ward [Wed, 31 May 2000 01:09:52 +0000 (01:09 +0000)]
Overhauled to expect 'self.extensions' (taken from 'ext_modules' in the
setup script) to be a list of Extension instances, rather than a list of of
(ext_name, build_info) tuples.  This is mostly a simplification, but
'check_extension_list()' got a lot more complicated because of the need to
convert the old-style tuples to Extension instances.

Temporarily dropped support for defining/undefining macros in the
'extensions' list -- I want to change the interface, but haven't yet made
the required changes in CCompiler and friends to support this nicely.

Also neatened up the code that merges 'extra_compile_flags' and the CFLAGS
environment variable.

24 years agoProvides the Extension class, a nicer way to describe C/C++ extensions than
Greg Ward [Wed, 31 May 2000 01:05:35 +0000 (01:05 +0000)]
Provides the Extension class, a nicer way to describe C/C++ extensions than
the old (ext_name, build_info) tuple.

24 years agoFixed typo noted by Guido Kollerie <gkoller@cs.vu.nl> in example code.
Fred Drake [Tue, 30 May 2000 14:39:45 +0000 (14:39 +0000)]
Fixed typo noted by Guido Kollerie <gkoller@cs.vu.nl> in example code.

24 years agoAdd 'r' to docstrings to avoid expanded backslash escapes. (Ka-Ping Yee)
Guido van Rossum [Tue, 30 May 2000 13:25:35 +0000 (13:25 +0000)]
Add 'r' to docstrings to avoid expanded backslash escapes.  (Ka-Ping Yee)

24 years agoAdded concrete example of when separate build and install might be
Greg Ward [Tue, 30 May 2000 03:00:43 +0000 (03:00 +0000)]
Added concrete example of when separate build and install might be
useful (suggested by Moshe Zadka).

24 years agoCosmetic tweak.
Greg Ward [Tue, 30 May 2000 02:04:54 +0000 (02:04 +0000)]
Cosmetic tweak.

24 years agoCatch CCompiler exceptions in 'setup()'.
Greg Ward [Tue, 30 May 2000 02:04:29 +0000 (02:04 +0000)]
Catch CCompiler exceptions in 'setup()'.

24 years agoImport exceptions from errors.py, not ccompiler.py.
Greg Ward [Tue, 30 May 2000 02:02:49 +0000 (02:02 +0000)]
Import exceptions from errors.py, not ccompiler.py.

24 years agoRemoved exceptions -- now in errors.py to avoid expensive import of ccompiler.
Greg Ward [Tue, 30 May 2000 02:02:48 +0000 (02:02 +0000)]
Removed exceptions -- now in errors.py to avoid expensive import of ccompiler.

24 years agoMoved the CCompiler exceptions here, to avoid having to import ccompiler.py
Greg Ward [Tue, 30 May 2000 02:02:14 +0000 (02:02 +0000)]
Moved the CCompiler exceptions here, to avoid having to import ccompiler.py
  just to get a little exception class.
No more string-based exceptions.

24 years agoChanged to catch compile/link failures and raise CompileError, LibError,
Greg Ward [Tue, 30 May 2000 01:56:44 +0000 (01:56 +0000)]
Changed to catch compile/link failures and raise CompileError, LibError,
or LinkError (exception classes defined in ccompiler.py).

24 years agoAdded 'dump_options()' for debugging output.
Greg Ward [Sun, 28 May 2000 23:54:00 +0000 (23:54 +0000)]
Added 'dump_options()' for debugging output.

24 years agoFactored '_set_command_options()' out of 'get_command_obj()'.
Greg Ward [Sun, 28 May 2000 23:53:06 +0000 (23:53 +0000)]
Factored '_set_command_options()' out of 'get_command_obj()'.
Added 'reinitialize_command()' -- lets us "push" option values in
  a controlled, safe way; this is a small change to the code, but
  a big change to the Distutils philosophy of passing option values
  around.  The preferred mode is still definitely to "pull" options
  from another command (eg. "install" fetches the base build directory
  from "build"), but it is now feasible to "push" options onto another
  command, when you know what's best for it.  One possible application
  will be a "config" command, which pokes around the system and pushes
  values (eg. include and library directories) onto the "build" command.
Added 'dump_option_dicts()' method (for debugging output).

24 years agoChanged order so 'clean' is right after the 'build' commands.
Greg Ward [Sun, 28 May 2000 23:49:03 +0000 (23:49 +0000)]
Changed order so 'clean' is right after the 'build' commands.

24 years agoOnly print debugging output if DEBUG (imported from distutils.core) is true.
Greg Ward [Sun, 28 May 2000 23:47:31 +0000 (23:47 +0000)]
Only print debugging output if DEBUG (imported from distutils.core) is true.

24 years agoMoved warnings out of 'finalize_options()' into 'run()'.
Greg Ward [Sun, 28 May 2000 23:47:00 +0000 (23:47 +0000)]
Moved warnings out of 'finalize_options()' into 'run()'.
Added a warning for 'bdist_base' directory.

24 years agoTypo fix in comment.
Greg Ward [Sun, 28 May 2000 20:29:48 +0000 (20:29 +0000)]
Typo fix in comment.

24 years agoSome far-reaching naming changes:
Greg Ward [Sat, 27 May 2000 17:27:23 +0000 (17:27 +0000)]
Some far-reaching naming changes:
  * Command method 'find_peer()' -> 'get_finalized_command()'
  * Command method 'run_peer()' -> 'run_command()'
Also deleted the 'get_command_option()' method from Command, and
  fixed the one place where it was used (in "bdist_dumb").

24 years agoArticle about 1.6 new features
Andrew M. Kuchling [Sat, 27 May 2000 11:28:26 +0000 (11:28 +0000)]
Article about 1.6 new features

24 years agoPatch from Andrew Kuchling: prune out the build and source distribution
Greg Ward [Sat, 27 May 2000 03:03:23 +0000 (03:03 +0000)]
Patch from Andrew Kuchling: prune out the build and source distribution
directories after all is said and done, so we don't accidentally include
those files in the source distribution.

(This is the quick and easy way to fix this; Andrew says: "Changing
findall() looked like it was going to be messy, so I tried this instead.
The only problem is that redundant directory traversals are being done,
walking through build/ only to throw out all the files found at the end.").

24 years agoSupport for the "install_headers" command: distribution option 'headers'
Greg Ward [Sat, 27 May 2000 01:36:14 +0000 (01:36 +0000)]
Support for the "install_headers" command: distribution option 'headers'
and method 'has_headers()'.

24 years ago'mkpath()' now detects non-string 'name' arguments -- this is a fairly common
Greg Ward [Sat, 27 May 2000 01:35:27 +0000 (01:35 +0000)]
'mkpath()' now detects non-string 'name' arguments -- this is a fairly common
bug when adding new code, so I thought I'd make it blow up earlier
than deep in posix.py.