Re: [vile] UTF-8/Latin9 problems

[Thomas Dickey]

On Thu, Jan 15, 2009 at 05:29:28PM +0100, Paul van Tilburg wrote:
> Hi!
> 
> So, I haven't tracked Vile's updates for a while.  I have updated my
> 9.6m install to 9.7j, but now, if I enter <compose>,",e to produce
> e-umlaut, but it produces \?EB, the latin1 varient of this character.
> My input charset is set to UTF-8, my LC_CTYPE is nl_NL.UTF-8 and
> file-encoding (or $buf-encoding) is set to UTF-8.
> Is this a regression?  Or did things change?

It's a regression - I just narrowed it down to one of the fixes for 9.7i,
as part of this item:

       + correct one place where inserted bytes were interpreted as 
         characters, which caused the map-command to insert UTF-8 bytes as 
         characters (report by Christopher Illies).

That touched two files, map.c and insert.c, and it seems the bug is in
the latter.  fwiw, my checkin comment for that says

   use lins_bytes() in inschar().  This seems to fix a case reported by
   Christopher Illies, where he created an insert map like this:
        :map! ^Ks, ^Vu0219
   and then did an insert of "^Ks,".  The result showed the UTF-8 bytes for
   the \u0219 (0xc8 and 0x99) inserted as separate characters.  That's because
   I used lins_chars() in this place.  The map worked from a script, but not
   interactively.  With this fix, both work.

(thanks for the bug-report).

-- 
Thomas E. Dickey <address@hidden>
http://invisible-island.net
ftp://invisible-island.net

signature.asc
Description: Digital signature