Bug#724115: hunspell: FTBFS: POD error

[Agustin Martin <agmartin@debian.org>]

On Tue, Nov 12, 2013 at 07:00:10PM +0100, gregor herrmann wrote:
> On Tue, 12 Nov 2013 18:34:52 +0100, Agustin Martin wrote:
> 
> > > > here's a patch for this issue.
> > Note that you are changing way more things that needed, including strings
> > that are explicitly in a given charset, which is wrong. 
> 
> That was on purpose ...

I mean that it included parts of the code, making it no longer work as
expected (yes, that is about old features). If script is changed to UTF-8
these unibyte strings need to be rewritten with escapes to stay unibyte
(hope this will work), not changed to UTF8.

>  
> > This can be fixed in a simpler way by declaring charset, see attached patch,
> 
> Right, for the bug itself "=encoding ISO8859-1" is enough.
> My experience is that pod2man with latin1 is not the best option,
> since it tends to transliterate stuff to ascii.
> 
> Let's try:
> 
> #v+
> % cat latin1.pod 
> =encoding ISO-8859-1
> 
> some nice Ümläüts
> #v-
> 
> % pod2man latin1.pod | grep "some nice"
> some nice U\*:mla\*:u\*:ts
> 
> Not bad; but when I run
>     % pod2man latin1.pod | grep "some nice" | man -l -
> I get
>     some nice Umlauts
> in the pager. Maybe because this is in an UTF-8 environment? Or
> something with the pager?
> 
> Anyway, that's why I prefer to have POD in UTF-8 and call pod2man
> with --utf8.

Thanks for the info. I later noticed pod2man misbehavior about this, but I
could not try harder. Also tried to unsuccessfully use POD escapes to
make it more portable. No luck.

I will have a look at this (I once wrote ispellaff2myspell). Now I think the
best is to change script to UTF8, but keep strings in code as escaped octal. 
Or rewrite that part.

Let me think about this. Hope to find time tomorrow.

Thanks for your feedback.

-- 
Agustin