|
|
http://scons.tigris.org/svn/scons/branches/core
........
r1970 | stevenknight | 2007-06-01 18:22:37 -0500 (Fri, 01 Jun 2007) | 4 lines
Import a vanilla Python 2.3 version of textwrap.py into the compatibility
library, so we can track the changes we'll make to it. (This isn't
actually used yet.)
........
r1971 | stevenknight | 2007-06-02 00:38:20 -0500 (Sat, 02 Jun 2007) | 2 lines
Add a compatibility module for the textwrap.py module introduced in Python 2.3.
........
r1972 | stevenknight | 2007-06-02 00:39:26 -0500 (Sat, 02 Jun 2007) | 2 lines
Remove spurious <para> tags.
........
r1973 | stevenknight | 2007-06-03 08:57:05 -0500 (Sun, 03 Jun 2007) | 2 lines
Improved help-text generation using a textwrap.TextWrapper object.
........
r1991 | stevenknight | 2007-06-10 16:03:18 -0500 (Sun, 10 Jun 2007) | 3 lines
Add compatibility versions of the all() and any() functions introduced
in Python 2.5.
........
r1992 | stevenknight | 2007-06-10 17:02:18 -0500 (Sun, 10 Jun 2007) | 8 lines
SCons-time portability fixes for Python 2.[12]:
-- Use "from __future__ import nested_scopes".
-- Create "False" and "True" builtins
-- Work around the lack of a "prefix =" keyword argument to the
Python 2.[12] version of the mktemp module.
-- Accomodate pickier single-element tuple syntax.
........
r1993 | stevenknight | 2007-06-10 17:27:43 -0500 (Sun, 10 Jun 2007) | 3 lines
Delay instantiation of pstat.Stats objects until after we override
sys.stdout, which as of Python 2.5 is captured when the object is created.
........
r1994 | stevenknight | 2007-06-10 21:22:42 -0500 (Sun, 10 Jun 2007) | 6 lines
Update various tests to handle the
File "SConstruct", line 1, in <module>
Messages in Python 2.5.
........
r1995 | stevenknight | 2007-06-10 21:32:16 -0500 (Sun, 10 Jun 2007) | 3 lines
Update tests to not raise strings as exceptions, which has been
deprecated in Python 2.5.
........
r1996 | stevenknight | 2007-06-10 21:41:57 -0500 (Sun, 10 Jun 2007) | 3 lines
Fix the Scanner hash unit test for Python 2.5. (Yes, it still works
on previous versions, too.)
........
r1997 | stevenknight | 2007-06-10 21:55:46 -0500 (Sun, 10 Jun 2007) | 3 lines
Make the mock Node object's side_effect attribute a list, so it's
iterable in Python 2.1 as well.
........
r1998 | stevenknight | 2007-06-10 22:04:26 -0500 (Sun, 10 Jun 2007) | 3 lines
Append an explicit tuple to the delayed_warnings list if there are
problems interpreting --debug=memoizer.
........
r1999 | stevenknight | 2007-06-11 11:09:07 -0500 (Mon, 11 Jun 2007) | 2 lines
Fix --debug=time with -j when no arguments are rebuilt (all up-to-date).
........
r2007 | stevenknight | 2007-06-14 13:56:35 -0500 (Thu, 14 Jun 2007) | 4 lines
Performance improvement when looking up Nodes: don't use is_String(),
just check for the initial '#' that specifies a top-relative lookup,
and handle the exceptions.
........
r2008 | stevenknight | 2007-06-14 16:57:47 -0500 (Thu, 14 Jun 2007) | 11 lines
First step in refactoring command-line flag processing: Split out the
current processing into its own module, with minimal additional changes.
Among the minimal changes:
-- Store delayed warnings (for deprecated --debug= keywords)
in the option parser object, not in a global variable.
-- Remove the OptParser variable itself from the SCons.Script globals.
It's going to change significantly (and no one's probably using
it anyway).
-- Don't move definition of the --version output with the OptParser,
keep it in Main.py.
........
r2009 | stevenknight | 2007-06-15 08:15:25 -0500 (Fri, 15 Jun 2007) | 3 lines
Refactor the test/explain.py script into three individual scripts
so it's easier to deal with.
........
r2010 | stevenknight | 2007-06-15 09:49:07 -0500 (Fri, 15 Jun 2007) | 3 lines
Handle Aliases in --debug=explain. This is kind of hard-coded for the
normal lookup, and should be better handled by the signature refactoring.
........
r2011 | stevenknight | 2007-06-15 17:25:37 -0500 (Fri, 15 Jun 2007) | 5 lines
Refactor use of the command-line parser object so it's localized to the
top-level main() function, and not passed down through _exec_main() or
to _main() itself. Replace its functionality with use of an exception to
signal that the top-level main() function should print its help message.
........
r2012 | stevenknight | 2007-06-17 23:34:26 -0500 (Sun, 17 Jun 2007) | 2 lines
Remove unnecessary import of __main__.
........
r2013 | stevenknight | 2007-06-17 23:48:06 -0500 (Sun, 17 Jun 2007) | 2 lines
Pass the options object to _main(), don't use a global.
........
r2014 | stevenknight | 2007-06-18 00:12:09 -0500 (Mon, 18 Jun 2007) | 6 lines
Qt test fixes for Windows:
Link against a static library created by the test infrastructure,
not a shared library.
Escape backslashes in Windows path names.
Skip test/QT/Tool.py if Qt isn't installed.
........
r2015 | stevenknight | 2007-06-18 10:46:17 -0500 (Mon, 18 Jun 2007) | 3 lines
Support GetOption('no_exec'), and update test/NodeOps.py to use it
instead of reaching into the SCons.Script.Main internals.
........
r2016 | stevenknight | 2007-06-18 11:04:39 -0500 (Mon, 18 Jun 2007) | 4 lines
Restore use of a global delayed_warnings variable so the chicken-and-egg
warning from trying to use --debug=memoizer on Python versions without
metaclasses has somewhere to go.
........
r2017 | stevenknight | 2007-06-18 11:37:59 -0500 (Mon, 18 Jun 2007) | 3 lines
Have the test infrastructure create a mock Qt shared library on UNIX,
static library on Windows.
........
r2018 | stevenknight | 2007-06-18 11:48:10 -0500 (Mon, 18 Jun 2007) | 2 lines
Pull more globals into the command-line parser options object.
........
r2023 | stevenknight | 2007-06-19 16:46:02 -0500 (Tue, 19 Jun 2007) | 3 lines
Refactor the __checkClass() and must_be_a_Dir() methods into a more
general and more efficient must_be_same() method.
........
r2025 | stevenknight | 2007-06-19 20:56:10 -0500 (Tue, 19 Jun 2007) | 3 lines
More clean up: change various self.fs.Entry() calls to calls through
the bound directory.Entry() method.
........
r2033 | stevenknight | 2007-06-20 20:03:23 -0500 (Wed, 20 Jun 2007) | 5 lines
The --debug=count option doesn't work when run with Python - O, or
from optimized compiled Python modules (*.pyo files), because the
counting is all within "#if __debug__:" blocks that get stripped.
Print a warning so it doesn't look like --debug=count is broken.
........
r2037 | stevenknight | 2007-06-21 10:42:40 -0500 (Thu, 21 Jun 2007) | 3 lines
Replace the _stripixes() function with a more efficient/readable version
that was checked in, but commented out, prior to 0.96.96.
........
r2040 | stevenknight | 2007-06-21 12:18:57 -0500 (Thu, 21 Jun 2007) | 2 lines
Ignore *.pyo files, too, since one of the tests now generates them.
........
r2051 | stevenknight | 2007-06-26 15:11:57 -0500 (Tue, 26 Jun 2007) | 5 lines
Arrange for graceful shutdown of the worker threads by writing None to
the requestQueue and having the worker threads terminate their processing
loops when they read it. We can then .join() the threads, to wait for
their termination, avoiding exceptions in the threading library module.
........
r2052 | stevenknight | 2007-06-26 15:12:53 -0500 (Tue, 26 Jun 2007) | 3 lines
Have the SWIG tests that use the Python.h header skip gracefully if
the Python development environment isn't installed.
........
r2053 | stevenknight | 2007-06-26 15:23:55 -0500 (Tue, 26 Jun 2007) | 3 lines
Massage the datestamp and IDs in the generated PDF so we can compare
before-and-after output reliably regardless of when generated.
........
r2054 | stevenknight | 2007-06-26 15:25:56 -0500 (Tue, 26 Jun 2007) | 3 lines
Fix the regular expression that matches the Qt warning message when
the moc executable is used as a hint.
........
r2055 | stevenknight | 2007-06-26 15:48:21 -0500 (Tue, 26 Jun 2007) | 2 lines
Fix 2.5.1 string exception warnings.
........
r2056 | stevenknight | 2007-06-26 19:23:22 -0500 (Tue, 26 Jun 2007) | 2 lines
Skip the scons-time tests if the Python version can't import __future__.
........
r2057 | stevenknight | 2007-06-26 22:11:04 -0500 (Tue, 26 Jun 2007) | 3 lines
Normalize PDF output in the bibtex rerun test as well. Commonize the
PDF normalization logic by putting it in QMTest/TestSCons.py.
........
r2058 | stevenknight | 2007-06-26 22:50:39 -0500 (Tue, 26 Jun 2007) | 3 lines
Duplicate a function declaration to suppress compiler warnings about a cast,
when using certain systems/compilers.
........
r2059 | stevenknight | 2007-06-26 22:53:12 -0500 (Tue, 26 Jun 2007) | 2 lines
Use the frtbegin when compiling Fortran programs using GCC 4.
........
r2060 | stevenknight | 2007-06-26 23:13:35 -0500 (Tue, 26 Jun 2007) | 2 lines
Make the object that goes into the shared library a shared object file.
........
r2061 | stevenknight | 2007-06-26 23:53:49 -0500 (Tue, 26 Jun 2007) | 4 lines
Split test/AS/AS.py into sub-tests for the live assemblers it tests.
Only test nasm for the known configuration of version 0.98* on a
32-bit x86 system.
........
r2063 | stevenknight | 2007-06-27 09:51:43 -0500 (Wed, 27 Jun 2007) | 2 lines
Fix searching for the rmic utility.
........
r2064 | stevenknight | 2007-06-27 10:26:42 -0500 (Wed, 27 Jun 2007) | 3 lines
Improved worker-thread termination in a separate Job.cleanup() method.
(Adam Simpkins)
........
r2087 | stevenknight | 2007-07-03 12:22:10 -0500 (Tue, 03 Jul 2007) | 7 lines
Get rid of unnecessary subclassing and complicating overriding of
__init__() and parse_args() methods in favor of more straightforward
initialization of the OptionParser object.
We may need to restore subclassing in the future, but if so we'll do it
in a more OO way.
........
r2088 | stevenknight | 2007-07-03 16:12:30 -0500 (Tue, 03 Jul 2007) | 2 lines
Fix a cleanup error (no self.p4d attribute) when Perforce isn't installed.
........
r2090 | stevenknight | 2007-07-04 03:23:57 -0500 (Wed, 04 Jul 2007) | 2 lines
Import the vanilla Python 2.5 optparse.py for use as a compatibility module.
........
r2091 | stevenknight | 2007-07-04 03:35:17 -0500 (Wed, 04 Jul 2007) | 5 lines
Use the new optparse compatibility module for command-line processing,
and remove the SCons/Optik/*.py modules, with appropriate subclassing
in Script/SConsOptions.py to preserve the way we print help text and
SCons error messages.
........
r2108 | stevenknight | 2007-07-08 22:57:08 -0500 (Sun, 08 Jul 2007) | 3 lines
Make all of the optparse.add_options calls more-or-less consistent in
how they call the keyword arguments.
........
r2109 | stevenknight | 2007-07-09 12:31:01 -0500 (Mon, 09 Jul 2007) | 6 lines
Consolidate command-line and {Get,Set}Option() processing and access
in a single subclass of the optparse.Values() class.
Allow all options, not just those that aren't SConscript-settable,
to set their default values when calling op.add_option().
........
r2110 | stevenknight | 2007-07-09 13:17:58 -0500 (Mon, 09 Jul 2007) | 4 lines
Handle initialization of command-line repository options by passing the
option arguments directly to the _SConstruct_exists() utility function,
not by setting a global variable.
........
r2111 | stevenknight | 2007-07-09 13:42:41 -0500 (Mon, 09 Jul 2007) | 2 lines
Remove the unused _varargs() utility function.
........
r2112 | stevenknight | 2007-07-09 15:21:51 -0500 (Mon, 09 Jul 2007) | 2 lines
Clean up how we use optparse (mainly for readability).
........
r2113 | stevenknight | 2007-07-10 15:50:08 -0500 (Tue, 10 Jul 2007) | 2 lines
More old-Python-version compatibility changes in optparse.py.
........
r2114 | stevenknight | 2007-07-10 16:46:42 -0500 (Tue, 10 Jul 2007) | 3 lines
Add support for a new AddOption() function to allow the SConscript
file(s) to define new command-line flags.
........
|
