Accepted mksh 59c-10 (source) into unstable
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA384
Format: 1.8
Date: Mon, 04 Oct 2021 02:44:17 +0200
Source: mksh
Architecture: source
Version: 59c-10
Distribution: unstable
Urgency: low
Maintainer: Thorsten Glaser <tg@mirbsd.de>
Changed-By: Thorsten Glaser <tg@mirbsd.de>
Changes:
mksh (59c-10) unstable; urgency=low
.
* Package CVS HEAD (again with locale tracking):
- [tg] Document MKSH_ASSUME_UTF8 will need to go away with full locale
tracking; (see Debian) locale will be federführend even in scripts…
- [tg] Correct several minor/cosmetic issues
- [tg] Add generator code for character classes, more consistent
- [tg] Use small stack buffers over allocating to reduce malloc pressure
in several more codepaths
- [tg] Add character class documentation to the manpage
- [tg] Small optimisations (less duplication; optimise for the common
codepath instead of the uncommon one) in code and testsuite
- [tg] EBCDIC: codepage consistency check, diagnostic; document more
- [tg] Plug a couple memory leaks; slightly increase performance
- [tg] Massive code and comment improvements during work on fixing the way
special characters (control characters, invalid UTF-8, etc.) are handled
or escaped, including emacs mode cursor positioning and output
- [tg] Reduce memory allocator pressure by using more on-stack buffers
- [tg] Improve outputting job information
- [tg] Harmonise some error reporting and use string pooling more
- [tg] Make rotating by 0 avoid C UB… *sigh*
- [tg, RT] Add new Build.sh option -A
- [RT, tg] Port to SCO OpenServer 3.2r4.2 and Xenix (crossbuilt for now)
- [tg] Fix FAQ2HTML.sh on præ-POSIX sh(1)
- [tg] Make testsuite not fail on rangeless tr(1) or symlinkless system
- [tg] Allow for <sys/select.h> substituting <sys/time.h> if both may not
be included at the same time (scosysv 3.2r4.2)
- [tg] Elide some configure-time tests on OEs where they are known
- [tg] Not only is fgrep(1) unportable, but so is grep(1) -F…
- [tg] Calculate the most negative number w/o triggering UB (I want Ͻ…)
- [tg] Improve signedness‑ and const-cleanliness of code
- [tg] Add strstr(3) and memmove(3) if missing
- [tg] Clear a few more tty flags on makeraw; inspired by ksh93
- [tg] Skip the rename builtin if rename(2) syscall is missing
- [tg] Add a signal handling abstraction; still require sigaction(2) in
the normal case, whitelist OSes with reliable signal(3) bsd_signal
- [tg, RT] Xenix has reliable signals but broken termios(4); cope
- [tg] Speed up configuring: skip a few checks for ancient compilers
- [RT, tg] Update and promote to working the port to Syllable Desktop;
improve the situation for LynxOS 3.0 and BeOS 6.2 “Zeta” and, slightly,
for BeOS 5.1 “Dano” and 5.0 “Maui”
- [tg] Check whether exec -a newargv0 prog args… works separately so it
isn’t caught through sh-mode-1 indirectly only; split that
- [tg] Begin switching to… slightly less insane… integer types
- [RT] Port to SkyOS (fragile), SerenityOS (begun), Minoca OS (good)
- [tg] Sanity-check some usual tools before using (build, test)
- [tg, RT, komh] Handle broken-but-extant getrusage(2)
- [tg] Make times(3) use work in the absence of CLK_TCK
- [tg] Actually act on getrusage(2) and times(3) errors
- [tg] When tabcompleting, set the mark to the beginning of the inserted
substring (the cursor is at or just one space past the end), similar
to how other editing commands behave; for xthechar on IRC
- [tg] When importing env vars, first match now wins, like in libc
- [tg] Exporting parameters to the environment now uses a stable order
- [tg] Move kill-region default binding to Esc+^W because ^W is usually
taken by WERASE, to unconfuse
- [RT, tg] Work around NeXTstep cc(1) more
- [tg] Testsuite: C.UTF-8 is now the default UTF-8 locale on all operating
systems except HP/UX where it still defaults to en_US.utf8 and Darwin
(Mac OSX) which uses just UTF-8 for some reason‽
- [tg] Run all tests with LANG=C since musl defaults to C.UTF-8 if all
locale parameters are unset (and talk about Debian glibc to do the same
exists and I’m considering this for MirBSD maybe as well)
- [tg] Make base-1 integers pass typeset -p reentry-safe
- [tg] Fix manual page misrendering a few Ns macros with GNU groff
- [tg] Uncouple HAVE_SETLOCALE_CTYPE=0 from (soon to be gone)
-DMKSH_ASSUME_UTF8={0|1}
- [RT] Make it easier to work around a broken $Config{perlpath}
- [tg] Use extended caret notation in bind input and output
- [tg] Prepare for moving raw octet mapping off U+EF80‥U+EFFF to
U-00200080‥U-002000FF but don’t actually do it until R60 release
- [tg] Use extended caret notation in the command line editor, etc.
- [tg] Restore transpose ability to work for nōn-UTF8 characters
- [tg] Make input handling more reliable, prepare some for 21-bit UCS
- [tg] Do not attempt to fcntl(2) on just-closed file descriptors…
- [tg] Warn if setting or clearing the close-on-exec flag fails
- [tg] Fix test for character device in mksh -T
- [tg] Address Coverity Scan checkers’ complaints
- [tg] Make casting around, EBCDIC, DEBUG builds more reliable
- [tg] Protect tab completion with extended caret notation escaping,
reported by lanodan on IRC
- [tg] Make Build.sh -g more useful-ish with other compilers and prefer
-Og over -O2 for optimisation then
- [tg] Revert turning on POSIX mode turns off UTF-8 mode from R56
- [tg] Advance to renaissance locale tracking; revert defining our
implementation-defined default locale to C and use the OS’
- [tg] Add ${var@^} for unambiguous extended caret notation
* Identify as “+Debian” vendor version; should’ve done that earlier…
* Bump Standards-Version, no change
Checksums-Sha1:
c6ac6d4a86fa755e57beb0ffeaec9667ed9cfdfd 2259 mksh_59c-10.dsc
d7e5a05bdfe437c864c39a3b720b3128b383dcb2 188900 mksh_59c-10.debian.tar.xz
Checksums-Sha256:
431187ec79760cbbd097b804aac6ff93fb923a7970086068dd8711eef743a36e 2259 mksh_59c-10.dsc
ec5e8ae4c208be6448281a8c5ca6035f096aa5c30031ae476a378db450ddac02 188900 mksh_59c-10.debian.tar.xz
Files:
c56756cdce98dfb3a76558e1b5b21b97 2259 shells optional mksh_59c-10.dsc
9848df6036b1572daa4067e811fd4a8e 188900 shells optional mksh_59c-10.debian.tar.xz
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (MirBSD)
Comment: ☃ ЦΤℱ—８ ☕☂☄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=wA/V
-----END PGP SIGNATURE-----
Reply to: