locale-related review, wcwidth() data

[Baptiste Daroussin]

5 déc. 2020 22:13:31 Yuri Pankov <yuripv at yuripv.dev>:

> Baptiste Daroussin wrote:
>> 5 déc. 2020 02:25:29 Thomas Munro <thomas.munro at gmail.com>:
>> On Sat, Dec 5, 2020 at 2:31 AM Baptiste Daroussin <bapt at freebsd.org> wrote:
>>>> I do like what I see here, the only reason I haven't review is that I can't
>>>> test, since the last modification from hrs@ in the locale generation tools each
>>>> time I try to regenerate the locales it fails.
>>>
>>> During install?  I noticed that too but wasn't sure of the correct fix, perhaps:
>>>
>>> diff --git a/tools/tools/locale/Makefile b/tools/tools/locale/Makefile
>>> index 76fff6acb17..b6ae2feadac 100644
>>> --- a/tools/tools/locale/Makefile
>>> +++ b/tools/tools/locale/Makefile
>>> @@ -95,7 +95,7 @@ install: install-${t}
>>> install-${t}:
>>>         cd ${LOCALESRCDIR}/${t} && \
>>>             rm -f Makefile *.src && \
>>> -           install -c ${t}/* ${LOCALESRCDIR}/${t}
>>> +           install -c ${.OBJDIR}/${t}/* ${LOCALESRCDIR}/${t}
>>> .  endif
>>> .endfor
>>
>> Nope that one was easy to figure out, but once the locales are regenerated, localdef dies on plenty of them (non unicode mostly).
>
> Now that you mentioned it, there's something I was thinking about for a long time now -- it's probably a sign that we need to mark all single-byte locales as set in stone, and stop regenerating the source for them as well as adding more and more workarounds against utf-8 charmap.  This is something I'm going to look into eventually as I believe it will make updating to new CLDR releases much easier, that is, if I'm not missing something here.
>
I fully agree!!!

Bapt