-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Format: 1.7
Date: Wed, 25 Apr 2007 11:36:42 +0000
Source: mksh
Binary: mksh
Architecture: source i386
Version: 29.2-1
Distribution: unstable
Urgency: low
Maintainer: Thorsten Glaser <tg@mirbsd.de>
Changed-By: Thorsten Glaser <tg@mirbsd.de>
Description:
mksh - enhanced version of the Korn shell
Changes:
mksh (29.2-1) unstable; urgency=low
.
* New upstream formal release; summary of changes:
+ a plethora of UTF-8 fixes:
- display control characters U+0080..U+009F the same as U+0000..U+001F,
i.e. a caret followed by the character corresponding to the control
character XOR 0x0040, treat their width as 2 subsequently
- fix crash (cpu hog in spinning loop) on meta-tab + backspace
- strip off UTF-8 byte order mark at beginning of input
- if a BOM is encountered, switch on utf-8 command line editing mode
+ in utf-8 command line editing mode, handle invalid input more strictly:
- if in x_literal(), i.e. the ^V mode (bind to quote), allow it as before
- if it's the start of an invalid multibyte sequence, reject with a beep
(e.g. if trying to input latin1 chars)
- if it's an invalid or partial multibyte sequence, reject silently
-> this makes command line editing much more robust
+ other bug fixes:
- in a rare condition (error path), the wrong function was used to copy
a string that could contain embedded NULs (encoded format), leading to
memory access past malloc'd area
- in the same path, fix an out-of-bounds access inherited from openbsd ksh
-> discovered on Debian GNU/Linux experimental ia64, glibc 2.5-beta
+ new functionality:
- if execve() fails, before passing the file to $EXECSHELL, open it and
try to parse a shebang line; if we find one, do it ourselves
(the good part of this is that it even works when there is a UTF-8 BOM
before the shebang magic sequence)
- for shebang processing, not only NUL or LF but also CR terminate the line
- enhancements to the "dot.mkshrc" sample file (which is now regularily
used upstream as well)
- if the internal function exchild() fails, don't just tell the user that
it failed, tell him WHAT failed (unless -DMKSH_SMALL)
+ code cleanup changes:
- remove unused functions, macros
- fix typos, errors, etc.
- shut up gcc 4.1.2 warnings
- Build.sh cygwin vs unix cleanup/simplification
- shrink manual page to 39 DIN A4 pages when output as postscript
+ reliability changes:
- if $CC supports -fstack-protector-all, add it to $CFLAGS
- if $CC supports -fno-tree-vrp, add it to $CFLAGS if $CC is subject to
the bug http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30785
- add mirtoconf check for "large file support", requested by bsiegert@,
needed for some *nix, idea and implementation hints from GNU autoconf
- add zsh workaround to Build.sh, just in case (untested)
* disable the possible workaround mentioned in the changes for 28.9.20070309
because I was unable to verify/test it; maybe it only applies to the glibc
in experimental anyway, we'll see to that later
* add a comment about the regression test needing openpty() to debian/rules
* remove non-ASCII (i.e. high-bit7) characters from diff/changelog
* slightly enhance package description
* properly indent homepage link in description, thanks KiBi (kfreebsd team)
Files:
fce5476662390cc35a32792c1573be7a 556 shells optional mksh_29.2-1.dsc
aef5820258d05a0d61a02f38d2e497b3 241062 shells optional mksh_29.2.orig.tar.gz
aaee008a0b83da48684dea66225d9130 11125 shells optional mksh_29.2-1.diff.gz
48148815c09d3b3f354249afff0defd5 159724 shells optional mksh_29.2-1_i386.deb
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iEYEARECAAYFAkYwNBUACgkQLkAIIn9ODhEz0gCfaduv+TBkw2aOqovb0fHnTQ7s
arYAnjuw/IKVOQ59ee+iVniUDz0806cz
=gyS3
-----END PGP SIGNATURE-----
Accepted:
mksh_29.2-1.diff.gz
to pool/main/m/mksh/mksh_29.2-1.diff.gz
mksh_29.2-1.dsc
to pool/main/m/mksh/mksh_29.2-1.dsc
mksh_29.2-1_i386.deb
to pool/main/m/mksh/mksh_29.2-1_i386.deb
mksh_29.2.orig.tar.gz
to pool/main/m/mksh/mksh_29.2.orig.tar.gz
