It seems that popups were intended to "grab" the mouse click that triggered them, but their intent was being lost. This commit does the necessary changes to let it happen and updates items that were trying to get advantage of it, because the semantics of `Control::grab_click_focus()` have changed a bit. Namely, it must be called **before** showing the modal.
This allows to popup a menu and activate an item in it in a single click-point-release cycle, instead of having to click once to open the menu and once more to pick an item.
This ability is extended even to context menus activated with the RMB (or any other mouse button, for that matter). The editor benefits from this in the context menu of the tree dock, which has been patched to opt-in for this feature.
This improves UX a bit by saving unnecessary clicks.
From now on, `PopupMenu` always grabs the click and also invalidates the first button release unless the mouse has moved (that's what `set_invalidate_click_until_motion()` was doing and now it's removed), so there is no longer the need of doing both things at every point a pop-up menu is shown.
fixes#17325.
The bone pose transform was created by setting the rotation and
**then** scaling the transform. This leads to object "deformation"
that's not intended.
Since create_outline can only make outline for PRIMITIVE_TRIANGLES,
when QuadMesh (which is PRIMITIVE_TRIANGLE_FAN) is used to create
outline, will leave `arrays` empty, and crash when it is being indexed
for "indices" subarray.
This PR shows error when there's only one surface and it is not
TRIANGLES. Also prevent the crash if it has more than one surface
and none of them are TRIANGLES (and any other cases that could leave
`arrays` empty) by checking the size of `arrays` == 8 before indexing
it, since the method seems to expect `arrays` to be of that size.
ProgressBar used the center size of the stylebox to calculate its minimum size, thus disallowing certain setups.
If the old behaviour is wanted, it can be forced by providing a custom minimum size, or by giving proper margins to the stylebox.
Fixes#17779.
They work exactly the same as current checkbox-decorated items, but in order to preserve compatibility, separate methods are used, like `add_radio_check_item()`. The other option would have been to add a new parameter at the end of `add_check_item()` and the like, but that would have forced callers to provide the defaults manually.
`is_item_checkable()`, `is_item_checked()` and `set_item_checked()` are used regardless the item is set to look as check box or radio button.
Keeping check in the name adds an additional clue about these facts.
Closes#13055.
When `RichTextLabel` calculated the actual width of columns, it let
them grow to sizes greater than its calculated `max_width`. Now this
PR ensures no columns grows beyond `max_width`, and the excess width
is distributed among the columns which can still grow.
It should fix#17731.
The hue variation matrix is hardcoded as transposed and thus the multiplication always yielded the original color value.
I wanted to keep the matrix and get the same result by switching the operands (getting a result as if the matrix was transposed).
Font update after resize relies on the viewport size which was updated
after the font was already refreshed, which resulted in artifacts when
it was rendered into the actual/new viewport size.
Fixes#15173.
Text overflowed canvas as tables didn't calculate correctly the width
of their columns. They used the whole table width available for each
column. Also, the `cell` parameter was wrongly parsed if used with its
optional argument (expand ratio).
This PR fixs the parsing of `cell` parameter (i.e. `cell=e`) and the
distribution of the full table width between columns, but it overrides
automatically the `expand` flag if the column is smaller than it could
be, to allow a better UX out-of-the-box. It keeps the `expand_ratio`
flag to let the user customize how every column grows in relation to
the rest.
Partially fix#6289.
It appears that some time ago users were supposed to be able to include the playback of sound effects in their animations by placing keys on the "playing" property. Back then the key frame editor took the value of the checkbox in the property_editor.
Somewhere / Sometime this behaviour changed and the key frame editor is now reading the actual value from the object instead of relying on the property editor.
This commit introduces a fake active field that is returned when reading the playing property in the editor. While the actual active flag is changed when playback is finished the fake one will stay the same thus allowing the user to take their time with setting the key in the animation editor.
Fixes#15459. When oversampling is enabled, glyphs may have fractional
size, but they are still rendered into integral pixels, which results in
them taking more space than was anticiped by autowrapping algorithm. The
solution here is to return ceiled width, which makes autowrapper
consider characters a bit larger than they are, but it doesn't hurt the
actual rendering and ensures there is enough space for the characters.
Make TileMap monitor its TileSet for changes and emit a signal when the TileSet changes. This makes the editor update and show the updated version of the TileSet.
Original code used a quick aproximation for simulating the
correspondent texel in the `TextureProgress` texture as radial
progress indicator. This lead to visualization errors. Changed it for
a Liang-Barsky line clipping algorithm stripped to its minimum for
this specific use case.
Fix#17364.
I had a grid container and tried to set rect.min_height larger in the
editor; that caused an infinite loop in GridContainer::_notification
at line 118. The reason is max_index was being set to the *height* of
the row, not the *index* of the row. So later when it tried to erase
that row and try again, there was nothing to erase.
I applied the same fix to the width code.
If you had a tree like Node2D->Sprite->Camera2D and you write a
code like $Node2D/Spr and chose the autocompletion sugested
Node2D/Sprite, the resulting string was $Node2D/Node2D/Sprite
instead $Node2D/Sprite. If you chose Node2D/Sprite/Camera2D, then
you ended with $Node2D/Node2D/Sprite/Camera2D.
Fix#15813.
playing speed of the animation being played.
As stated in #16550, there is no way to get the playing speed of an
animation. The `playback_speed` property is for the whole AnimationPlayer
node, and the argument passed to `play` method is another scale applied
in top of the AnimationPlayer properties.
Thus, the actual playing speed is AnimationPlayer.speed_scale *
AnimationPlayer.playback.current.speed_scale. If it is not playing, the
method returns 0.
It is possible to try to add an invalid object as a navmesh through
GDScript which results in an engine crash. This creates a debug message
that should help the user figure out what's wrong.
Previously this option seemed to be the sole responsible for enabling
physics processing in Viewport, while several other features like
tooltips and debugging collision hints rely on it.
All this logic is moved to internal processing (it's incorrect to let
it be affected by users disabling physics/idle processing), and disabling
physics object picking no longer affects the internal physics processing.
Fixes#17001.
- Editor font hinting can now be tweaked in the Editor Settings.
- DynamicFonts used in projects now have tweakable hinting settings
in their DynamicFontData child. Changes will be visible upon
reloading the scene in the editor.
The previous logic with VERSION_MKSTRING was a bit unwieldy, so there were
several places hardcoding their own variant of the version string, potentially
with bugs (e.g. forgetting the patch number when defined).
The new logic defines:
- VERSION_BRANCH, the main 'major.minor' version (e.g. 3.1)
- VERSION_NUMBER, which can be 'major.minor' or 'major.minor.patch',
depending on whether the latter is defined (e.g. 3.1.4)
- VERSION_FULL_CONFIG, which contains the version status (e.g. stable)
and the module-specific suffix (e.g. mono)
- VERSION_FULL_BUILD, same as above but with build/reference name
(e.g. official, custom_build, mageia, etc.)
Note: Slight change here, as the previous format had the build name
*before* the module-specific suffix; now it's after
- VERSION_FULL_NAME, same as before, so VERSION_FULL_BUILD prefixed
with "Godot v" for readability
Bugs fixed thanks to that:
- Export templates version matching now properly takes VERSION_PATCH
into account by relying on VERSION_FULL_CONFIG.
- ClassDB hash no longer takes the build name into account, but limits
itself to VERSION_FULL_CONFIG (build name is cosmetic, not relevant
for the API hash).
- Docs XML no longer hardcode the VERSION_STATUS, this was annoying.
- Small cleanup in Windows .rc file thanks to new macros.
Found via `codespell -q 3 --skip="./thirdparty,./editor/translations" -I ../godot-word-whitelist.txt`
Whitelist consists of:
```
ang
doubleclick
lod
nd
que
te
unselect
```
- The input handling is done into several distinct functions, and the
code is more consistent.
- The actions' history is more precise ("Edited CanvasItem"
is now "Rotated CanvasItem","Moved CanvasItem",etc...)
- Fixed a little bug about input key events not forwarded correctly to plugins
- IK is followed by default when you move a bone node, the alt-key allow
you to move it normally
This behavior better matches other gui toolkits. A selected disabled
button still can't be interacted with but it can now be selected. This
seems to be what QT and GTK do also.
This fixes#16131
if the audio player is set to play again due to the order of calls in
_notification. First it emits the signal, and later it disable the internal
processing regardless what the callback did.
Changed to emit the signal at the end to ensure the changes done at callback
remains.
