Re: GNU sed version 4.2.1: on OS X, C locale gets aliased to UTF-8

[Max Horn]

Hi again!


On 06.09.2012, at 11:08, Paul Eggert wrote:

> Alas, I don't use OS X, or Mac OS X, or whatever it's called
> these days, so I can't easily test any fix, and I'd like some
> discussion if possible.

Sure. 

> 
> In January Bruno wrote that my proposed patch
> <http://lists.gnu.org/archive/html/bug-gnulib/2012-01/msg00319.html>
> would break gettext-based i18n of GNU programs on OS X; see
> <http://lists.gnu.org/archive/html/bug-gnulib/2012-01/msg00342.html>.

And I refuted those claims in 
<http://lists.gnu.org/archive/html/bug-gnulib/2012-06/msg00013.html>


> 
> In July Max Horn proposed a different patch, which he said is more logical
> <http://lists.gnu.org/archive/html/bug-gnulib/2012-07/msg00097.html>.

This is incorrect -- indeed, if you read the link you just gave carefully, 
you'll see that I proposed the exact same patch you proposed back in January. 
Moreover, in the email I sent last night, I once more proposed that patch

The different patch you are talking about was proposed by yourself :-), see 
<http://lists.gnu.org/archive/html/bug-gnulib/2012-06/msg00015.html>. And I 
later confirmed that this also fixes the issue in my reply to your email. 
However, I don't really see why that patch is better or more logical... Rather, 
I find the original trivial patch much more logical :-).


> Max's patch would set the codeset to be ASCII if
> the codeset is UTF-8 but MB_CUR_MAX is 1.  This seems like
> it should do the trick, and shouldn't break any system
> that isn't already broken.
> 
> Nobody has replied to Max's proposal, as far as I know.
> Are there any objections if I install his July patch?

Well, I still would prefer the original patch. There were some concerns about 
it, but I tried to address and refute every single one. So far replied to that 
either... In particular, I think Bruno's analysys that this would break 
gettext-based i18n of GNU programs on OS X is incorrect, see once more 
<http://lists.gnu.org/archive/html/bug-gnulib/2012-06/msg00013.html>.  In 
addition to my explanations there, let me add that I (and everybody who has 
installed GNU sed via Fink) has been using the proposed fix for months, and I 
can assure you that i18n is working fine.

So, I still would very much prefer the original patch, which simply maps 
US-ASCII to ASCII. But in the end, I don't care so much how gnulib fixes it, as 
long as it gets fixed in some way... :)


Cheers,
Max