Differences to the plain GNU version in the packages collection:
* We do not install the shell wrapper "yacc" (it is supplied because
POSIX requires it and we already have a yacc command).
Changes since 1.75:
* Numerous bug fixes and improvements including:
+ Compatibility (with 1.35 and Solaris yacc) changes;
+ Fixes for GCC 3.2.1;
+ Use Yacc style of conflict reports;
+ Fix bug where error locations were not being recorded correctly;
+ Fix bad interaction with flex 2.5.23.
Please see the ChangeLog file supplied with the bison source code
for more details.
*) Use "close-on-exec" semantic on internal file descriptors if
underlying platform supports this feature. This makes sure the file
descriptors are closed by the kernel upon execution of exec(3) by
the application.
Changes :
* Factored the help-formatting code out of OptionParser into
some new classes (HelpFormatter and subclasses) in help.py. This
should make it a lot easier to customize how help is formatted.
* Added the notion of "option groups": an OptionParser can now
contain several option groups, each which contains several options.
The main purpose of this is to enable sensibly-grouped help output,
but it opens up all sorts of interesting (and largely untested)
possibilities for code to throw whole option groups around instead
of individual options. Added two new classes: OptionGroup, and
OptionContainer for code common to OptionParser and OptionGroup.
(OptionContainer should be invisible to programmers using Optik).
* Added the 'description' attribute and set_description() method to
both OptionParser and OptionGroup (actually OptionContainer).
Again, this is to make help output more useful.
* Made it easier for OptionParser subclasses to decide whether
they should have the standard "help" option, by moving the logic
from class level to the _populate_option_list() method.
* Added the "choice" option type, which is just a string type
constrained to a fixed set of values.
* Added method get_default_values() to OptionParser.
* Rewrote how OptionParser recognizes abbreviated long
options; removed a redundant internal instance attribute.
* Simplify parsing logic in OptionParser a tad by relocating a loop
and renaming _process_arg() to _process_args().
Changes :
Version 0.8.12
* Add --isliteral option to warn when using is/is not against literal
(e.g., XXX is 5 YYY is not 'abcdef', etc)
* Add --modulo1 option to warn when using (expr % 1),
which is constant when expr is an integer or string
* Add -4/--noeffect option to disable warnings for 'Statement has no effect'
* self is not first argument warning can be disabled with -S/--self ''
* Don't print duplicate warnings when importing pychecker
* Fix other spurious warnings when importing pychecker
* Fix bug for spurious invalid arguments when passing a dict inline
* Fix bug for spurious Statement has no effect when print >> x, ... ,
* Add @option_file command line argument to read options from a file
(used on platforms which can't have too many options, ie Windows)
* Fix several crashes
* Line numbers should be close for Python 2.3 and if run on optimized code
Version 0.8.11
* Improve error message for syntax errors from user files
* Fix pychecker.bat so it should work now
* Add a warning for using __coerce__ in new-style classes
* Add --deprecated option for using deprecated modules or functions
* Add a warning for using functions with security problems (os.t[e]mpnam)
* Add a warning for comparing against True/False or defining True/False
* Add --badexcept option to warn when using string exceptions or
classes not derived from Exception to raise/catch exceptions
* Fix spurious warnings from using (test and 'true' or 'false)
DESCR:
The Library of Assorted Spiffy Things (LibAST - formerly libmej) was
originally designed as a companion library to Eterm. However, it
contains many features that other projects may find useful, so it
has been broken out into its own library. Features include thorough
debugging and memory tracking engines, various string manipulation
functions, and a flexible config file parsing engine.
Changes:
* Add new role ATK_ROLE_APPLICATION [Bill Haneman]
* Correct implementation of i18n support [Padraig O'Briain]
* Add signal "active-descendant-changed" to AtkObject [Padraig]
* Add relation types ATK_RELATION_EMBEDS and ATK_RELATION_EMBEDS_BY
[Bill Haneman]
* Add atk_hyperlink_is_inline() [Bill]
* Add new functions atk_object_add_relationship, atk_object_remove_relationship,
atk_role_get_localized_name, atk_action_get_localized_name; add new roles,
states and relationships; add text-attributes-changed signal for AtkText
interface. [Padraig O'Briain]
* Add po directory and update atk_role_get_localized_name() to allow it to
return localized name [Padraig]
* Add implementtaion of atk_object_factory_get_accessible_type() for
AtkNoOpObjectFactory [Padraig]
* Add ATK_LAYER_WINDOW {Mark Mc Loughlin]
and some translation updates and bug fixes.
Changes:
* Improve font solection for Hangul tone marks
* Improve range definitions for basic-xft [Jungshik Shin]
* Draw unknown character boxes with X primitives for the
X backend instead of hoping for a suitable glyph [Morten Welinder]
* For Xft/FT2, give an informative error mesage when no fonts are found.
* Support shape modules with ranges beyond the BMP [Federic Zhang]
* Use octal escapes rather than literal UTF-8 [Arnaud Charlet]
* Use g_[sf]printf where appropriate [Matthias Clasen]
* Fix the Arabic-words-broken-by-underlines problem by
stripping out non-shape affecting attributes then adding them
back after shaping.
* Hangul shaper fixes [Changwoo Ryu, Jungshik Shin]
* Handle compatability ideographs in basic-x shaper [Federic Zhang]
* Implement pango_ft2_font_map_set_default_substitute() [Matthias Clasen]
* Hangul module bug fixes and improvements [Changwoo Ryu, Jungshik Shin]
- Move rendering of precomposed characters to hangul-xft shaper
- Handle Hangul tone marks
* Add a new wrap mode for PangoLayout - PANGO_WRAP_WORD_CHAR [Alex Larsson]
* Indic module improvements [Eric Mader]
* Fixes to generic OpenType handling [Eric]
* Improve language specific charset ordering basic-x11 [James Su, Brian Yuan]
* Some merges of fixes by Werner Lemberg to the original
FreeType OpenType code.
* Robustify and otherwise improve pango_color_parse() [Matthias Clasen]
* Some cross-compiling fixes [J. Ali Harlow]
* Merge a lot of the code between the Xft and FT2 backend,
so that the FT2 backend gets the recent improvements to Xft.
* Add proper language tag support to the Xft/FT2 backends.
* Various effienciency fixes for the Xft/FT2 backends.
* Support for version 2 of the Xft library. [Keith Packard]
* Convert the freetype backend over to using fontconfig,
instead of MiniXft for font cataloging.
* A port of the Indic OpenType code from ICU to Pango. [Eric Mader]
* A new hangul-xft [Changwoo Ryu]
* Various cleanups to the PangoXft and PangoFT2 public APIs.
... as well as the usual bug fixes.
Major changes:
* Improve generation of pseudo-random integers [Morten Welinder,
Sebastian Wilhelmi]
* Avoid literal UTF-8 in favor of octal escapes [Owen, Tomas Ogren]
* Add copy of the Trio library to build and use for printf() when
system printf isn't good enough. Add g_printf()/etc. [Matthias Clasen]
* Add g_str_has_suffix()/g_str_has_prefix() [Alex Larsson]
* Add g_markup_parse_context_get_element() [Matthias]
* Add g_utf8_strreverse [Matthias]
* Add g_ascii_strtoull() [Tim Janik]
* Support scanning of 64-bit values with GScanner [Tim]
* Add g_set/get_application_name() [Havoc Pennington]
* Add G_LIKELY()/G_UNLIKELY() macros for hinting branch probabilities.
Use for g_return_if_fail(). [Matthias Clasen]
* Add G_GNUC_DEPRECATED macro [Tom Tromey]
* Improve the seeding algorithm of GRandom to avoid problems
with certain pathological seeds. Support G_RANDOM_VERSION=2.0
environment variable. [Sebastian Wilhelmi]
* Improve thread configure checks, use -pthread where applicable
[Sebastian]
* Improve handlng of thread priorities [Sebastian]
...as well as many bug fixes and translation updates.
