Re: [screen-devel] Bug in rendering terminal application

[Amadeusz Sławiński]

Hi,

yes there are problems with how screen handles utf-8, rewriting
encoding handling to be more manageable is still on my todo list.

I already experimented a bit here 
https://github.com/amade/screen/tree/encoding-rewrite
but ultimately it needs spending quite some time on it and thinking how
to handle verious stuff. For example there needs to be clean separation
between strings that need to be char* and those that can be uint32_t*.
And there is probably few more things that I don't remember right now.

Ultimately it would be great if we also got some automated testsuite
for how screen prints various characters. (If there are any takers it
would be great ;) ).

Amadeusz

On Tue, 26 Jun 2018 22:31:55 +0200
Maarten van Malland <address@hidden> wrote:

> I've been digging around myself regarding this bug, and it seems the
> problem is slightly different from what I previously thought. In the
> source code of that application there are some special characters
> being used as text. In a normal utf-8 terminal this is displayed as:
> 
> 𝗚𝗔𝝡𝗘 𝗢𝗩𝗘𝗥!
> 
> In ISO-8859-2 this is displayed as:
> 
> đ<9d><97><9a>đ<9d><97><94>đ<9d><9d>Ąđ<9d><97><98>
> đ<9d><97>˘đ<9d><97>Šđ<9d><97><98>đ<9d><97>Ľ!"
> 
> However, in the master branch of Screen using utf-8 it is is displaying:
> 
> 헚흡 헢험!
> 
> Well, that's weird! For some reason Screen interprets that code as
> Korean characters. Funny thing is that google translates it back as
> "Take a breath!"; that's exactly what I'm doing right now ;).
> 
> 
> On Mon, Jun 25, 2018 at 9:47 PM Maarten van Malland
> <address@hidden> wrote:
> >
> > I think I've encountered a bug in the master branch with a specific
> > terminal application: https://github.com/0xbsec/terminal2048 This
> > application is working fine in several other terminals (termite, st,
> > terminator) as well as within tmux, but for some reason the fonts are
> > not rendered correctly within screen v5. In screen v4 it's not working
> > at all, so my suspicion is that this application is using truecolor
> > and screen v5 is having some issue with it. Any idea what might be
> > going on here?  
>