bug-libunistring
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [bug-libunistring] Unicode 15.1.0


From: Simon Josefsson
Subject: Re: [bug-libunistring] Unicode 15.1.0
Date: Fri, 27 Oct 2023 09:43:13 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)

Bruno Haible <bruno@clisp.org> writes:

>> I'm sure people will come along asking for libidn2 to support some of
>> the new characters, so an update of libunistring would be great.
>
> Does libidn2 use libunistring as a package, or does it import part of it
> from gnulib directly? That is, do you wish to have a new libunistring release
> soon?

Libidn2 uses libunistring if available, otherwise it falls back to
internally provided gnulib modules.  Most deployment uses libunistring,
I believe, but a new release is not important for me: if the gnulib
modules are updated to Unicode 15.1.0 I can make use of them and test
that Unicode 15.1.0 characters works inside libidn2 and tell people to
test.  This gives us time to fix any problem too before a new
libunistring release happens.

>> Is the
>> process to update libunistring to a new Unicode release documented
>> anywhere?  I suppose some things are not mechanic, but some of the steps
>> are.
>
> It's documented in the sense that we have ChangeLog entries from the
> previous upgrades. Typically the tasks are like this:
>   1. Understand what were the main changes in that Unicode version.
>   2. Download the data files.
>   3. Run gnulib/lib/gen-uni-tables.c against the new data files,
>      fixing aborts on the go.
>   4. With these modified gnulib files, build libunistring and run its
>      test suite.
>   5. Update files that implement UAX annexes (e.g. line breaking).
>   6. Update version number metainfo.
>
> I'll probably get to it in the next few days; at a first glance, the
> changes don't appear to be large.

Yeah it looked small, but it is hard to really tell from the Unicode
changelog: what would be needed is a diff of all specifications, and
going through all modifications manually to verify that it doesn't
modify implementation.  While possible, I tend to prefer relying on test
suites to catch anomalies.  I think Unicode publishes test vectors too,
how much of them does libunistring uses?  Unfortunately libidn2 doesn't
support the latest Unicode IDN test vectors because they changed format,
otherwise we would use them too.

/Simon

Attachment: signature.asc
Description: PGP signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]