LGTM, but it seems the patch is misformatted. Did you use git to
format (and send) the email?
If it's too much of a hassle, please feel free to open a PR on the
mirror: github.org/gioui/gio.

Sorry for the trouble,
Elias

Lovely, thank you! I've merged your change and the deployment
should happen automatically.

Elias

On Sat, 19 Apr 2025 at 07:33, Oliver Lowe <o@olowe.co> wrote:
>
> The previous layout was missing the seconds.
>
> Signed-off-by: Oliver Lowe <o@olowe.co>
> ---
> I was writing a little RSS and Atom feed package and used feeds I'm
> subscribed to for testing, then noticed parsing the gioui.org feed
> failed.

I forgot to write to you yesterday. LGTM, merged.

Elias

On Wed, 9 Apr 2025 at 05:17, Dave Akers <dave@dazoe.net> wrote:
>
> Signed-off-by: Dave Akers <dave@dazoe.net>
> ---
>
> I'm not 100% about this patch. Using seat.pointerFocus requires setting seat.pointerFocus to nil. Pointer frame event fires after pointer leave which will stop window.flushScroll and window.flushFling from being called. I don't have a touch device to test if this causes any problems or not.
>
>
>  app/os_wayland.go | 27 +++++++++++++++------------
>  1 file changed, 15 insertions(+), 12 deletions(-)

On Mon, 7 Apr 2025 at 20:17, Dave Akers <dave@dazoe.net> wrote:
>
> Signed-off-by: Dave Akers <dave@dazoe.net>
> ---
>  app/os_wayland.go | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/app/os_wayland.go b/app/os_wayland.go
> index e68d3e13..32a8dec0 100644
> --- a/app/os_wayland.go
> +++ b/app/os_wayland.go
> @@ -1172,6 +1172,9 @@ func (w *window) updateCursor() {
>         if ptr == nil {
>                 return

On Fri, 4 Apr 2025 at 06:46, Dave Akers <dave@dazoe.net> wrote:
>
> Hello,
>   I've been experimenting with gio on wayland. Until a few weeks ago I've never looked at the wayland protocol and the wayland code in gio is the first wayland code I've seen. Tracking down and fixing a few of the pointer related bugs has caused another. If the pointer moves from an editor across a small border or space then leaves the window there is a panic because the pointer leave event sets the app.window's seat field to nil, then because the cursor was being changed there is a call to updateCursor which tries to read a field from window.seat. A simple nil check would fix it but then I looked deeper, and I have questions and maybe I'm missing something.
> Why is the seat field set to nil? I couldn't find anywhere it is checked.
> Why does the window struct have the seat field anyway? The window struct has a disp field which is the wlDisplay which has a seat field, why not use that instead?
>

The field was added in 5ce1e982827ba0ab4af963b8719701aab8681ac9 by
Pierre (CC'ed). It's been some time since it was merged, but it seems
to me the window.seat field tracks whether the window has the pointer
focus (seat.pointerFocus is set to the window at the same time w.seat
is set). I believe that's because calls such as
set_pointer/move/resize are only appropriate when the window has the

Please add a note describing the purpose of the new field, ideally with a link
to documentation.

Please also sign off your patch as described in the contribution guide:
https://gioui.org/doc/contribute. At the same time, use the Go convention
for patch subjects, e.g.

app: [Wayland] pass correct serial to wl_pointer_set_cursor

...

Thanks,
Elias

Thank you! I've merged your patch with an additional fix to window.setCursor.

Elias

Good question. I've raised https://todo.sr.ht/~eliasnaur/gio/641 to
discuss a fix to this problem.

Elias

Thanks! Merged.

Elias

LGTM, but I must ask you to sign off your change, see

https://gioui.org/doc/contribute

Elias