Re: localisation of 'info' not working

[Benno Schulenberg]

Karl Berry wrote:
>>http://lists.gnu.org/archive/html/bug-texinfo/2006-08/msg00001.html
>> The two links in that mail are broken, 
>
> I guess LFS managed to break their archive url's.  I couldn't
> find them either.  Maybe Alexander could tell us.

Alexander answered and pointed to this one instead:
http://linuxfromscratch.org/pipermail/lfs-dev/2005-August/052945.html
The screenshots are linked to at the bottom.

Compare for example:
http://linuxfromscratch.org/pipermail/lfs-dev/attachments/20050816/82fbbbb6/attachment.png
http://linuxfromscratch.org/pipermail/lfs-dev/attachments/20050816/82fbbbb6/attachment-0001.png
The linebreaking hyphen in "well-formed" is getting mangled, and 
other lines that contain a hyphen are inappropriately wrapped.

>     texinfo should in my opinion be removed
>     from the Translation Project
>
> That would be extremely discouraging.

It would.  I was trying to apply some pressure.  :)

> Even on that path, there 
> is still makeinfo and install-info ...

But those programs are hardly ever used by a normal user, but mostly 
only by package maintainers and distributors.  So localising those 
has relatively little value.  If 'info' itself isn't localised, 
there's little point in localising the other two.

> I guess I agree both
>
> (a) with your implication that occasional multibyte character
> breakage (especially in man output) isn't worth disabling
> translation altogether (I did not realize this was the
> implication; for that matter, UTF-8 was not as ubiqitious even a
> year+ ago as it has become), and also
>
> (b) your second suggestion:
>
>     or only when showing man pages?
>
> sounds doable enough, although I don't know any way to test it.

Alexander's reply to this was:
"No. This is too slippery a slope. This will just give people the 
impression that it is OK to leave small multibyte-related bugs, 
with no definition of "small" (and it is small only because nobody 
really uses texinfo with non-Latin-alphabet texts). Please fix 
counting of multibyte characters when wrapping lines and when 
displaying prompt,and don't break lines in the middle of a 
multibyte character."

> Actually figuring out how to handle UTF-8 in Info is beyond me. 
> If someone else wants to tackle it, that would be great.

Sergey?

> I've put it on the TODO list.

One thing that could be done now is add a FIXME to the comment on 
line 157 in info.c and mention that this a brute-force workaround, 
to avoid some misformatting of man-pages.  Seeing a bug documented 
in the program itself is better than having to dig out a changelog 
entry.  But much rather I'd see the workaround removed completely.  
Let people see the misformatting and let them complain and clamour 
for repair.

Benno