Re: SDL terminal type

On Sat, 22 Aug 2026 at 22:31, Eli Zaretskii <[email protected]> wrote:

>
> "C-x v h" is your friend, if you want to know.  Find the dates when
> this text was added or modified, and then look for related discussions
> on Emacs forums.>


The relevant additions to PROBLEMS have been made on May 27, 2022 and June
2, 2025. I tried to look for messages on emacs-devel around these dates,
but couldn't find anything applicable to this issue. What are the other
Emacs forums I should check?


> Which is why the text says "might".  I see no reason to describe all
> the details of the situations where this happens, because that doesn't
> help users when they bump into this.
>

The word "might" makes this even blurrier. But that diverts from the point
I was trying to make, which is that there are two distinct cases here:

1. The Input Method (IM) mechanism "steals" the key combination C-S-u. This
is the only key combination affected by IM, unless the input method is
configured to do exotic things.

2. The rest of the problems are, as I see it, caused by the fact that the
PGTK port does not handle key events in the same way as the other ports.
These problems do not go away by disabling the "GTK input methods", be it
either using the pgtk-use-im-context function or ibus-setup. There is not
much the user can do to fix this other than hack the PGTK C source code.

Disabling the input methods, e.g. using the pgtk-use-im-context function
resolves the problem described in case nr 1, but only that. It *does* help
the user to describe these details, because that helps the user which
problem he/she is trying to solve: that the C-S-u does funny things, or
that C-S-u is misreported as C-u. The text in etc/PROBLEMS conflates these
two cases, which makes me suspect the author has not fully taught through
the problems.

It is still implemented in GTK in this case, isn't it?  Then "GTK
> feature" is not really misleading, is it?
>

The GTK implementation is limited to GTK consulting ibus (or some other IM
framework) on how to handle key input before passing the (possibly
translated) input to the application. Ibus is an external software
component, not a part of Emacs, and the mechanism is not limited to GTK.
The problem just happens to manifest itself on the PGTK build, which is
built on GTK.

We will not remove anything, but if you want to suggest another
> workaround for these problems, please do: additions in that directions
> are always welcome.
>

Maybe the part about the Unicode "feature" could be useful for explaining
what the C-S-u issue is about. I would not recommend disabling ibus without
a clear explanation of what the implications of this action is, just to
salvage one key combination. The same thing can be done with the
pgtk-use-im-context function, without disturbing other apps on the machine.

Now, the C-> being reported as > remains a mystery to me, I have not been
able to reproduce that. Maybe it is dependent on the keyboard layout?