Re: Would there be a drawback of using the same graphical toolkit on every platform?

[Arthur Miller <arthur.miller@live.com>]

<tomas@tuxteam.de> writes:

> On Sun, Feb 20, 2022 at 06:53:45PM +0100, Arthur Miller wrote:
>
> [...]
>
>> I opposed to the rhetorics of Po who is blaming it on Gtk bugs and devs having
>> their own ideas on input [...]
>
> Now we're in blame territory. The "close X display thing" is actually a
> long standing Gtk bug (around Gtk 2). X programs could always connect to
> more than one display, and could close one or more of them. Gtk chose
> to just crash the program when closing a display. This was acknowledged
> as a bug, but never fixed.
>

It was the only one concrete mentioned, but there were also words like "various"
and some other statments on devs having strange ideas, not working well on
Windows, while it seems to work fine for many applications etc.

> Have a look here [1] for that long winded history.

Thank you for the link.

> Now you can describe calling this "a bug" as rhethoric, but to me, a
> spade is a spade is a spade. Don't get me wrong: the Gtk people are
> in their full right to WONTFIX this, of course.

 I don't see it as a "wontfix", it looks to me like they
are just asking for more comprehensible stack traces:

> Emacs calls abort() when closing the display to deal with something that GTK 2 did when coupled with Emacs's own client/server architecture where any process can suddenly become the owner of a display connection. People keep posting useless stack traces that end with the intentional abort() call inside Emacs, useless comments about "this still happens to me" (yes: Emacs still calls abort()), or any other random Emacs stack trace coming from any other random issue they have.

> GDK could use somebody going through the code with a smaller test case than the whole of Emacs, to check if closing a display connection on different backends breaks the existing API contract. If that happens:

>    open a new bug
>    track the stack traces and see where we're trying to access a display post-closure
>    possibly open a merge request that fixes all the instances

> Until that time, I'm going to close this issue because it's ridiculous.

Quoted from: https://gitlab.gnome.org/GNOME/gtk/-/issues/221