Re: Unicode versioning in glib

[Simon Josefsson <jas extundo com>]

Owen Taylor <otaylor redhat com> writes:

> Simon Josefsson <jas extundo com> writes:
>
>> I'm looking for a library that supports unicode normalization
>> (decomposition) and I discovered that glib supports this (are there
>> other LGPL or GPL libraries for this?) which is good.  The problem is
>> that I'm implementing some protocols which requires that the unicode
>> version (and in particular decomposition tables) must be of a certain
>> unicode version (3.2).
>> 
>> Is there any interest in supporting multiple unicode versions (for
>> decomposition or in general) in GLIB?  I assume that Gnome in general
>> would want to use the latest available unicode version, but this won't
>> work for me.
>> 
>> I haven't tested, but it seems I could simply borrow
>> gen-unicode-tables.pl and generate the tables which I need in my
>> project, which is probably what I will start with.  But if there is
>> interest in supporting this in GLIB, it is probably better.
>
> We aren't going to support multiple versions of Unicode in GLib,
> no. The amount of memory/disk space spent on the Unicode tables
> is already quite significant without having to deal with multiple
> copies.
>
> (Not to mention that it would significantly complicate the API 
> to have variants of functions taking Unicode version numbers.)

OK, I understand.  Thanks for the answer.

> I believe that the Unicode consortium has sufficient attention to
> backwards compatiblity to make versioned Unicode support not
> needed in typical usages.

Then my uses aren't typical, I guess.  Most if not all currently
pursued IETF standards require, for interoperability and security
reasons, that you use Unicode version 3.2 and no other version.