[Patch]: Ncurses frame drawing

[Micha Nelissen]

Corinna Vinschen wrote:

> Erm... I was still mulling over this code since I thought there's
> something wrong.  It took some time until it occured to me that this
> implementation overrides the original value of current_codepage, if
> the application accidentally happens to send \E[11m twice.  That
> shouldn't be possible.
> 
> I'd prefer if the value of original_codepage is set to the same value
> as current_codepage in environ.cc (codepage_init).  It should not
> be manipulated in fhandler_console.cc (char_command).

While thinking about the problem to solve, I now took a slightly 
different approach. See, those ncurses actually need to 'disable' 
translation from ansi to oem because that screws up the frame 
characters. So instead of original_codepage variable, I introduced a 
bool alternate_charset_active that states if an alternate charset is 
active. If so, it disables translation. Works perfectly for me and I 
think is a cleaner approach. No messing with remembering / initializing 
  / faking codepages.

Regards,

Micha.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: charset.patch
URL: <http://cygwin.com/pipermail/cygwin-patches/attachments/20031014/49b8bd1e/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: charset.log
URL: <http://cygwin.com/pipermail/cygwin-patches/attachments/20031014/49b8bd1e/attachment.log>