githubmirror/libidn2

NEWS
543 строки19 KB
Simon Josefsson
version 2.3.8
08 мар 2025, 23:51
08 мар 2025, 23:519bc3ac7
Libidn2 NEWS -- History of user-visible changes.                -*- outline -*-

* Noteworthy changes in release 2.3.8 (2025-03-08) [stable]

** Unicode 15.1.0 table updates.
Now U+19DA is DISALLOWED again (see version 2.3.4 release notes).

** The release tarball is now reproducible.

** We publish a minimal source-only tarball generated by 'git archive'.

** The release tarball uses tar --format=ustar.

** The idn2 tool now binds the "gnulib" domain for translations.

** Update gnulib files and various build/maintenance fixes.

* Noteworthy changes in release 2.3.7 (2024-01-27) [stable]

** Really include tests/standalone.sh in tarball.

* Noteworthy changes in release 2.3.6 (2024-01-27) [stable]

** Bump libtool version numbers to reflect API/ABI addition.

** Include tests/standalone.sh in tarball.

* Noteworthy changes in release 2.3.5 (2024-01-27) [stable]

** Declaration of future API/ABI backwards compatibility stability.
GNU libc dlopen libidn2 and use the name libidn2.so.0 for this.  We
believe that it will be too challenging to ever do hard ABI break that
for normal libraries is justified to remove deprecated APIs.  Thus we
decided that we will support the current ABI for a long time.  Of
course, if really convincing arguments for doing a ABI break appears
in the future we may re-consider, but take this as a declaration of
intent of will and that future ABI breaks should be discussed and
co-ordinated with the glibc team first.

** Add public APIs for raw Punycode encoding/decoding.
Normal applications rarely need this, but it cleans up the code and
allow for external testing of the APIs, and resolve
<https://gitlab.com/libidn/libidn2/-/issues/80> due to earlier use of
weak symbols for internal symbols _idn2_punycode_encode and
_idn2_punycode_decode.  We will support these internal symbols for
backwards compatibility.  This allows a clean migration path for code
that is still using the internal names.

** Bump required gettext version to 0.19.8 for musl-libc.
Reported by Helmut Grohne in
<https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=999510>.

** Un-deprecate idn2_to_ascii_4i and make it NUL terminate output.
The API idn2_to_ascii_4i was deprecated in version 2.1.1 released in
2019-02-08.  In that release, the API was also modified to not
NUL-terminate the output.  That is contrary to the old libidn2
behaviour, the behaviour of libidn's API idna_to_ascii_4i, and the API
documentation for the function.  Since we are not likely to ever break
backwards API/ABI compatibility in libidn2, and the deprecated gaurds
leads to some trouble (see report in
<https://gitlab.com/libidn/libidn2/-/merge_requests/93>) we decided to
un-deprecate this function, as supporting it is not costly and the
majority of code that cares about conformance has likely been
modified.  This will fix the error code and NUL termination report by
Yegor Bychin in <https://gitlab.com/libidn/libidn2/-/issues/100>.  We
still encourage you to use the replacement API/ABI idn2_to_ascii_4i2
instead, when appropriate.

** Compiler warning improvements.
As before, compiler warnings are enabled by default.  You may disable
them using ./configure --disable-gcc-warnings or turn them into fatal
errors using ./configure --enable-gcc-warnings=error to add -Werror
and sensible -Wno-error='s.  Based on gnulib's manywarnings, see
<https://www.gnu.org/software/gnulib//manual/html_node/manywarnings.html>.

** tests: Added script tests/standalone.sh suitable for integrators.
The main purpose is to test a system-installed libidn2 library and
idn2 tool, suitable for distributor checking (a'la Debian's
autopkgtest/debci).  It may also be used to test a newly built libidn2
outside the usual 'make check' infrastructure.  To check that your
system libidn2 library and idn2 tool is working, invoke the script
with `srcdir` as an environment variable indicating where it can be
find the source code for libidn2's tests/ directory (it will use the
directory name where the script is by default):

tests/standalone.sh

If your system libidn2 is too old to pass certain tests, disable them
using STANDALONE_DISABLE like this:

STANDALONE_DISABLE='*punycode*' tests/standalone.sh

See the script for more parameters.  If the libidn2 under testing is
too old and has known bugs, that should cause tests to fail, which is
intentional.

** Various minor build fixes and translation updates.

** API and ABI is backwards compatible with the previous version.
idn2_punycode_decode: ADD.
idn2_punycode_encode: ADD.

* Noteworthy changes in release 2.3.4 (2022-10-23) [stable]

** Support for Unicode 15.0.0.  Closes: #112.
We now uses Unicode.org's IDNA2008 tables rather than IANA's.  See
<https://gitlab.com/libidn/libidn2/-/issues/112> and
<https://lists.gnu.org/archive/html/help-libidn/2022-10/msg00000.html>
for rationale, which can be summarized into 1) IANA are still on
2019-era Unicode version 12 and we wish to support Unicode version
12-15, 2) consistency with some other implementations, 3) the only
incompatibility related to U+19DA is deemed to have minor real-world
consequences.  Thus we break backwards compatibility for U+19DA in
this release compared against libidn2 0.11..2.3.3 thus reverting back
to the libidn2 <= 0.11 behaviour.  We decided to not bump ABI version
and believe this is the best choice going forward as well for minor
internal non-API related ABI changes.

** Gnulib updated and now libunistring-optional is used.
This allows you to force libidn2 to use internal libunistring with the
following command: ./configure --with-included-libunistring

* Noteworthy changes in release 2.3.3 (2022-07-11) [stable]

** Upgrade IDNA Tables from Unicode 11 to 12.

** Upgrade TR46 Tables from Unicode 13 to 14.

** Updated gnulib files and various build fixes.
Gnulib's Unicode code claims conformance to Unicode 14.0.0 rather than
Unicode 9.0.0.  A bug in Libidn2's build system was fixed that caused
the system libunistring to be used even though the system version was
too old.

** Self-check improvements.
Self-checks for the idn2 command line tool were added.  Closes: #96.
The C self-checks in tests/ should now be usable outside of the
libidn2 build environment, for system integration checks of a
system-installed libidn2.

* Noteworthy changes in release 2.3.2 (2021-07-19) [stable]

** Upgrade TR46 tables from Unicode 11 to Unicode 13.
Now U+32FF works.  Fixes
<https://gitlab.com/libidn/libidn2/-/issues/77>.

** Fix build errors related to doc/idn2--help.texi.
With older makeinfo, this would manifest itself by looking for
idn2-help.texi instead.  Further, some BSD grep implementations did
not like the \+ regexp command, so we now rely on the (apparently)
more portable * regexp command.  Reported-By: David Gessel.  See
<https://gitlab.com/libidn/libidn2/-/issues/93>.

** doc: Improve GTK-DOC manual.

** doc: Don't ship HTML/PDF manual.  Drop custom CSS stylesheet.

* Noteworthy changes in release 2.3.1 (2021-05-12) [stable]

** Implement full roundtrip for lookup functionality
With TR64 enabled (default), 'â˜ºi' was converted to 'xn-- o-oia59s'.
The output contains an illegal space and thus could not be decoded any
more.  Reported-by: Chris Malton.  See
<https://gitlab.com/libidn/libidn2/issues/78> and
<https://gitlab.isc.org/isc-projects/bind9/issues/1610>.

** Fix domain too long error

** doc: idn2.1 and libidn2.texi automatically get idn2 --help output.

** Updated gnulib files and various build fixes.
In particular, it no longer attempts to detect a host CC compiler.
See <https://gitlab.com/libidn/libidn2/-/issues/83>.

* Noteworthy changes in release 2.3.0 (2019-11-14) [stable]

** Mitre has assigned CVE-2019-12290 which was fixed by
   the roundtrip feature introduced in 2.2.0 (commit 241e8f48)

** Update the data tables from Unicode 6.3.0 to Unicode 11.0

** Turn _idn2_punycode_encode, _idn2_punycode_decode
   into compat symbols (Fixes #74)


* Noteworthy changes in release 2.2.0 (2019-05-23) [stable]

** Perform A-Label roundtrip for lookup functions by default

** Stricter check of input to punycode decoder

** Fix punycode decoding with no ASCII chars but given delimiter

** Fix 'idn2 --no-tr64' (was a no-op)

** Allow _ as a basic code point in domain labels

** Fail building documentatino if 'ronn' isn't installed

** git tag changed to reflect https://semver.org/


* Noteworthy changes in release 2.1.1 (2019-02-08) [beta]

** Revert SONAME bump from release 2.1.0

** Fix NULL dereference in idn2_register_u8() and idn2_register_ul()

** Fix free of random value in idn2_to_ascii_4i()

** Improved fuzzer (which found the above issues)

** Fix printf() crash in test-lookup.c on Solaris

** Check for valid unicode input in punycode encoder

** Avoid excessive CPU usage in punycode encoding with
   large inputs

** Deprecate idn2_to_ascii_4i() in favor of idn2_to_ascii_4i2()

** Restrict output length of idn2_to_ascii_4i() to 63 bytes


* Noteworthy changes in release 2.1.0 (2019-01-04) [beta]

** Two exposed functions are no longer exposed:
   _idn2_punycode_encode() and _idn2_punycode_decode() which were
   meant to be used internally only. The output needs additional
   checks to be used safely.
   This is the reason to for the SONAME bump, just in case.

** Fix label length check for idn2_register_u8()

** Remove compiler warnings

** Use gnulib-python tool for bootstrapping if possible

** Improve build system (several small issues)

** Add missing error messages to idn2_strerror_name()

** Improve docs and remove typos

** Update gnulib


* Noteworthy changes in release 2.0.5 (2018-05-18) [beta]

** Switched the default library behavior to IDNA2008 as amended
   by TR#46 (non-transitional). That default behavior is enabled when
   no flags are specified to function calls. Applications can utilize the
   %IDN2_NO_TR46 flag to switch to the unamended IDNA2008. This is done in
   the interest of interoperability based on the fact that this is what application
   writers care about rather than strict compliance with a particular protocol.

** Fixed memleak in idn2_to_unicode_8zlz().

** Return error (IDN2_ICONV_FAIL) on charset conversion errors.

** Fixed issue with STD3 rules applying in non-transitional
   TR46 mode.

** idn2: added option --usestd3asciirules.


* Noteworthy changes in release 2.0.4 (2017-08-30) [beta]

** Fix integer overflow in bidi.c/_isBidi()

** Fix integer overflow in puny_decode.c/decode_digit()

** Improve docs

** Fix idna_free() to idn_free()

** Update fuzzer corpora


* Noteworthy changes in release 2.0.3 (2017-07-24) [beta]

** %IDN2_USE_STD3_ASCII_RULES disabled by default.
 Previously we were eliminating non-STD3 characters from domain strings
 such as _443._tcp.example.com, or IPs 1.2.3.4/24 provided to libidn2
 functions. That was an unexpected regression for applications switching
 from libidn and thus it is no longer applied by default.
 Use %IDN2_USE_STD3_ASCII_RULES to enable that behavior again.

** Fix several documentation issues

** Fix build issues

** Modernize gtk-doc build infrastructure.


* Noteworthy changes in release 2.0.2 (2017-04-27) [beta]

** Fix TR46 transitional mode

** Fix build issue on OSX

** Fix several documentation issues

* Noteworthy changes in release 2.0.1 (2017-04-22) [beta]

** idn2 utility now using IDNA2008 + TR46 by default

** Several doc fixes

* Noteworthy changes in release 2.0.0 (2017-03-29) [beta]

** Version numbering scheme changed

** Added to ASCII conversion functions corresponding to libidn1 functions:
- idn2_to_ascii_4i
- idn2_to_ascii_4z
- idn2_to_ascii_8z
- idn2_to_ascii_lz

** Added to unicode conversion functions corresponding to libidn1 functions:
- idn2_to_unicode_8z4z
- idn2_to_unicode_4z4z
- idn2_to_unicode_44i
- idn2_to_unicode_8z8z
- idn2_to_unicode_8zlz
- idn2_to_unicode_lzlz

** The idn2 manual page is generated from markdown text instead of utilizing
** help2man on the generated tool.

** Including idn2.h will provide libidn1 compatibility functions
** unless IDN2_SKIP_LIBIDN_COMPAT is defined. That allows converting
** applications from libidn1 (which offers IDNA2003) to libidn2 (which
** offers IDNA2008) by replacing idna.h to idn2.h in the applications'
** source.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.16 (2017-01-16) [alpha]

** build: Fix idn2_cmd.h build rule.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.15 (2017-01-14) [alpha]

** Fix out-of-bounds read.

** Fix NFC input conversion (regression).

** Shrink TR46 static mapping data.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.14 (2016-12-30) [alpha]

** build: Fix gentr46map build.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.13 (2016-12-29) [alpha]

** build: Doesn't download external files during build.

** doc: Clarify license.

** build: Generate ChangeLog file properly.

** doc: API documentation related to TR46 flags.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.12 (2016-12-26) [alpha]

** All changes by Tim Rühsen <tim.ruehsen@gmx.de> except stated otherwise.

** Builds/links with libunistring.

** Fix two possible crashes with unchecked NULL pointers.

** Memleak fix.
Reported by Hanno Böck <hanno@hboeck.de>.

** Binary search for codepoints in tables.

** Do not taint output variable on error in idn2_register_u8().

** Do not taint output variable on error in idn2_lookup_u8().

** Update to Unicode 6.3.0 IDNA tables.

** Add TR46 / UTS#46 support to API and idn2 utility.

** Add NFC quick check.

** Add make target 'check-coverage' for test coverage report.

** Add tests to increase test code coverage.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.11 (2016-11-03) [alpha]

** Fix stack underflow in 'idn2' command line tool.
Reported by Hanno Böck <hanno@hboeck.de>.

** Fix gdoc script to fix texinfo syntax error.

** Fix build failure of self-tests on platforms without version scripts.
Reported by Dagobert Michelsen <dam@opencsw.org>.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.10 (2014-06-25) [alpha]

** Update gnulib files.

** Dual-license the library.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.9 (2013-07-23) [alpha]

** Fix broken IANA link.
Apparently IANA does not provide persistent URLs to their registries.

** Fix automake bootstrap issue.

** Update gnulib files.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.8 (2011-09-28) [alpha]

** idn2: Fix build warnings.
Reported by Didier Raboud <odyx@debian.org> in
<https://bugs.debian.org/643427>.

** Update gnulib files.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.7 (2011-08-11) [alpha]

** libidn2: Fix missing strchrnul and strverscmp uses.
Reported by Ray Satiro <raysatiro@yahoo.com>.

** Update gnulib files.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.6 (2011-05-25) [alpha]

** tests: Use -no-install instead of -static to fix --disable-static.
Reported by Robert Scheck <robert@fedoraproject.org>.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.5 (2011-05-18) [alpha]

** Fix NFC check to compare entire strings.
Some non-NFC strings were permitted when they should have been
rejected.  Reported by Robert Scheck <robert@fedoraproject.org>.

** Self tests are not run under valgrind by default anymore.
Use --enable-valgrind-tests if you want to run self tests under
valgrind.  The reason was that there were too many false positives on
some platforms with valgrind issues in system libraries.  Self tests
are still run under valgrind by default when building from version
controlled sources.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.4 (2011-05-06) [alpha]

** libidn2: Fix domain name maximum size issue.
Domain names in string representation can be 254 characters long if
they end with a period, or 253 characters long if they don't end with
a period.  The code got this wrong and used 255 characters all the
time.  The documentation for the IDN2_DOMAIN_MAX_LENGTH constant is
improved.  We now pass two more of the IdnaTest.txt test vectors.
Reported by "Abdulrahman I. ALGhadir" <aghadir@citc.gov.sa> and
explanation from Markus Scherer <mscherer@google.com>.

** tests: Added several new Arabic test vectors.
From "Abdulrahman I. ALGhadir" <aghadir@citc.gov.sa>.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.3 (2011-04-20) [alpha]

** doc: Added Texinfo manual.

** doc: Added man pages for all API functions.

** examples: Added examples/lookup and examples/register as demo.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.2 (2011-03-30) [alpha]

** Added command line tool "idn2".

** Added more test vectors from Unicode.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.1 (2011-03-29) [alpha]

** IDNA2008 Lookup+Register functions are now operational.
The implementation is still subject to changes, and thus no API/ABI
stability guarantees are made.  We are now inviting comments both on
the API (as before) but also on the actual behaviour.  Any unexpected
outputs are from here on considered as real bugs.

** API and ABI is backwards compatible with the previous version.

* Noteworthy changes in release 0.0 (2011-03-09) [alpha]

** Initial draft release for public review of the API.
IDNA2008-Lookup is fully implemented except for 1) the optional
round-trip conversion part, and 2) the context rules are not
implemented.  IDNA2008-Register is not yet implemented.  The
implementation is known to be sub-optimal and ugly, please review the
interface and ignore the code!  Several changes are planned in the
internal implementation.

----------------------------------------------------------------------
Copyright (C) 2011-2025 Simon Josefsson
Copyright (C) 2018-2025 Tim Ruehsen

This file is free software: you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or (at your
option) any later version.

This file is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
General Public License for more details.

You should have received a copy of the GNU General Public License
along with this file.  If not, see <http://www.gnu.org/licenses/>.