Re: [Qemu-devel] [PATCH 0/5] curses: wide character support

[Samuel Thibault]

Hello,

Gerd Hoffmann, on Wed 26 Oct 2016 12:05:55 +0200, wrote:
> Back online now after being sick for a while.

Oh, sorry about that, hoping you're fine.

> > Samuel Thibault, on Sat 15 Oct 2016 21:53:03 +0200, wrote:
> > > This patch series adds wide character support to the curses frontend of 
> > > qemu,
> > > thus allowing to fix a lot of input and output issues with e.g. accented 
> > > letters
> > > and semi-graphic glyphs. Since qemu can't know the encoding of the VGA 
> > > font, the
> > > user has to specify it (just like he has to specify the keyboard layout 
> > > with
> > > -k). I used option -f to make it simple for now, but I welcome any other 
> > > idea :)
> > 
> > I forgot to mention that I updated the patches according to the reviews
> > on the list (assume cursesw support, and use -display option).
> 
> Good, looks much better now without all the #ifdefs.
> 
> /me wonders whenever we should do the same with iconv.  It's part of the
> POSIX.1-2001 specs, and we have 2016 now.  Do we *really* need configure
> checks and #ifdefs for it?

I would only say that libiconv is not always installed by default on BSD
systems for instance.

> Also we could use iconv for the cp437-to-unicode mapping instead of
> having a hard-coded table for it.

Ah, indeed.

> Given how close the freeze deadline is now I'm tempted to cherry-pick
> patches 0-3 and prepare a pull request ASAP.

You mean patches 1-3? Patch 5 would be very useful too to fix typing accented
letters, and it doesn't depend on patch 4.

Thanks,
Samuel