emacs-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Bugfix for utf-8 XTerm/MinTTY and (set-input-meta-mode t)


From: Max Mikhanosha
Subject: Re: Bugfix for utf-8 XTerm/MinTTY and (set-input-meta-mode t)
Date: Wed, 02 Jun 2021 10:21:26 +0000

On Tuesday, June 1st, 2021 at 4:06 PM, Stefan Monnier 
<monnier@iro.umontreal.ca> wrote:

> > Both XTerm and MinTTY, when configured to send meta modifier as 8th
> > bit while in utf-8 mode, will first add 8th bit, and then encode
> > resulting character with utf-8. For example Meta-X is encoded
> > as ?x+128 = #248 codepoint, encoded as 0xc3,0xb8
>
> How did they end up with that weird design?

It seems to be logical extension to preserve backward compatibility as much
as possible. I mean since without UTF-8, in meta-as-8th-bit mode, Meta-x 
generates #248,
just having terminal in UTF-8 mode should not change that. And in UTF8, starting
sequences that have 8th bit set indicate start of the encoding, so sending 
128-256
as is would be confused for UTF8 sequence by the receiver. Obvious solution 
would be
to just UTF-8 encode the output that non-utf-8 terminal would be sending.

> I mean they could have made meta toggle the 24th bit, for example, so it
> doesn't collide with other existing characters.

There is XTerm solution for this, called modifyOtherKey resource with new enum, 
which can be set
so that any modifiers even on ordinary keys like M-x would generate properly 
structured
ESC[ sequences describing the modifiers. I agree that in perfect world we would 
have come out
with some binary bitmask solution, rather than current thing where terminal can 
send you 20 byte
sequence for Ctrl-Alt-Shift-PageDown, but it is what it is.

> This design is quite weird since it breaks all the latin-1 chars of
> unicode plus all the uses of meta with non-ASCII chars.
>
> How do they encode M-λ ?
> Is it also sent as the same byte-sequence as `?λ + 128 = ?л` ?

Unfortunately yes, at least mintty in meta-as-8th bit mode (which is my 
terminal on cygwin) just
dumbly shoves 8th bit into even wide characters (like when I press Alt+letter 
in a cyrillic layout), but
my patch does not change that.  TBH Xterm maybe smarter, and just generates 
English M-x when you press
M-x when on a different keyboard layout, or you can probably make it behave 
like this with some xkb
config magic. Proper way to support multi-modifier key sequences is by  using 
modifyOtherKeys:2 but it would
need to have wider adoption than just xterm, hopefully it will percolate to 
other terminals just as
xterm-direct truecolor mode.

For now i'm pretty happy with meta as 8th bit mode, as it allows me to use 
stuff like M-C-v that is sent
as a single char (or 2 in utf8 mode), and with a bit of magic M-S-v and such 
work too, so all my keybindings
are exactly the same regardless if I'm in a terminal or GUI frame.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]