Unable to load right_ptr from the cursor theme

[tekstryder]

• Arch Linux | Kernel 6.7.11
• Gnome-shell | Mutter 46.0
• Wayland (meson_options: xwayland=false)
• Geany 2.0
• Gtk 3.24.41

Seeing these errors since Gnome 46.

Mar 30 08:00:22 geany[10386]: Unable to load right_ptr from the cursor theme
Mar 30 08:00:22 geany[10386]: g_object_unref: assertion 'G_IS_OBJECT (object)' failed
Mar 30 12:52:13 geany[15506]: Unable to load right_ptr from the cursor theme
Mar 30 12:52:13 geany[15506]: g_object_unref: assertion 'G_IS_OBJECT (object)' failed

For context, see the relevant issues/merges/discussions for the whole Adwaita cursor kerfuffle this past dev cycle.

• https://gitlab.gnome.org/GNOME/adwaita-icon-theme/-/issues/278
• https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3295
• https://gitlab.gnome.org/GNOME/adwaita-icon-theme/-/commit/74e9b79471236320d2af4925d6c5bb7df22380ce
• https://gitlab.gnome.org/GNOME/adwaita-icon-theme/-/merge_requests/65

[elextr]

Well don't use dead tortoise as your theme is likely the only answer Geany can give, we don't don't ship or set icon themes.

[tekstryder]

we don't don't ship or set icon themes

No, but you utilize cursors.

I suppose this is more a request for Geany supporting CSS xcursor standards[1][2] than finding icon themes that support esoteric cursors.

I'm not sure what the intended use of right_ptr is by Geany. What is that supposed to indicate to the user?

[1] https://www.w3.org/TR/css-ui-4/#cursor
[2] https://developer.mozilla.org/en-US/docs/Web/CSS/cursor

[b4n]

I'll have to check the details from the OP, but all custom cursors are probably coming from Scintilla. IIRC it uses a right pointed cursor over the left margin. Le 31 mars 2024 13:54:22 GMT+02:00, tekstryder ***@***.***> a écrit :

…

> we don't don't ship or set icon themes No, but you utilize cursors. I suppose this is more a request for Geany supporting CSS xcursor standards[1][2] than finding icon themes that support esoteric cursors. I'm not sure what the intended use of `right_ptr` is by Geany. What is that supposed to indicate to the user? [1] https://www.w3.org/TR/css-ui-4/#cursor [2] https://developer.mozilla.org/en-US/docs/Web/CSS/cursor -- Reply to this email directly or view it on GitHub: #3809 (comment) You are receiving this because you are subscribed to this thread. Message ID: ***@***.***>

[tekstryder]

IIRC it uses a right pointed cursor over the left margin

Ah, you're right (pun intended). Still unclear what that intends to indicate, but that's where the right_ptr attempt occurs.

[elextr]

Yes, Scintilla can use the cursors here and it just asks GDK for them.

It is documented that a right cursor is shown over margins and Geany does not set it, just leaves it as default. A PR making a setting to select a different cursor would probably be accepted so long as default was as is.

Elsewhere Scintilla notes use of another cursor icon for clickable links, not used by Geany, but I think plugin(s) might use it.

A quick look at the discussion the OP linked seems to suggest that the problem is being fixed by linking under the old names as well as the new names, so I doubt its worth it doing anything in Scintilla or Geany. The problem should only affect bleeding edge rolling distros for a short time, and if it doesn't get in a release should not appear in stable distros.

[tekstryder]

Thank you for looking into this and your through explanation!

if it doesn't get in a release should not appear in stable distros.

No, that "compromise" merged during the past dev cycle, and was included in the tagged release. All rolling/current distros that default to Adwaita are missing right_ptr.

There'd be cursor chaos currently among apps, I'd imagine, had they moved to strictly/only supporting CSS cursor names as was the intent of the original change this cycle.

Not sure this issue warrants a separate option, as the default cursor is simply displayed over margins, and this can only really be 'seen' by the system journal error. Cool to know you'd be open to a PR tho!

[elextr]

if it doesn't get in a release should not appear in stable distros.

No, that "compromise" merged during the past dev cycle, and was included in the tagged release. All rolling/current distros that default to Adwaita are missing right_ptr.

Sorry, if the linking of the old names compromise was in the release why is right_ptr missing, or did they forget that one?

[tekstryder]

why is right_ptr missing, or did they forget that one?

Possibly, but there are other more hotly debated one's missing/undecided too, such as left_ptr_watch and various dnd.

I only know enough to see that right-ptr was axed in the initial change, which was later partially reverted by what I referred to as a 'compromise'... albeit sans the right_ptr.

I'm not a gnome/gtk affiliate, just a curious user following the fallout from this change, so I've filed issues with a few affected applications such as here, Evolution Mail, and Foot terminal. The Foot developer wasn't too keen on this change either but did make some quick accommodations for preferring CSS cursor names.

I really never knew how many folks were so passionate about cursors! I don't care so much about my pointer's style, but I do cherish a clean system journal.

[elextr]

I really never knew how many folks were so passionate about cursors!

I'm not sure its so much about cursors as it is G* folks going back on their word and breaking GTK3 applications again, just proves that we can't trust 'em, Leopards never change their spots etc etc.

Oh well, my answer is "Don't use dead tortoises as themes, make a proper one you lazy ...." 😁

Basically a "default" theme is never going to be right for anyone, so it should be as lousy as possible (like default HTML) which Adwaita achieves IMNSHO so all distros should make their own and not be lazy [end rant]

I don't know why Scintilla uses different cursors, possibly to match the behaviour of the initial Windows control it is based on. It tries to be consistent on all platforms.

Perhaps Scintilla should switch to checking the return here and and falling back to default if fail.