Revert the changes from PR #1044 and #1045 and standardize on Object ** encoding in ptrcall#1123
Conversation
|
We need tests for this issue (#1045 (comment))
I think we should merge your PR #1101 asap. |
33605e1 to
87a9e4c
Compare
|
I've updated the tests so that they fail due to the regressions from #1044 and #1045. (So, this failure is a success!) Also, the tests seemed to have stopped failing on the bugs that #1044 and #1045 originally set out to fix! It looks like we were no longer getting a ptrcall in the situation we needed to, so I also swapped out the The tests do pass when paired with the Godot PR godotengine/godot#77410 |
87a9e4c to
7336564
Compare
Object ** encoding in ptrcall
|
The crash happen at overried |
Yes. If you use this PR together with Godot PR godotengine/godot#77410 then that crash won't happen. In a personal project I have a working |
…standardize on `Object **` encoding in ptrcall
7336564 to
ad72601
Compare
Object ** encoding in ptrcallObject ** encoding in ptrcall
|
Now that Godot PR godotengine/godot#77410 has been merged, I've rebased this and taken it out of draft. The CI should pass now that the Godot changes are in and a new artifact has been created. If not, I'd like to get that fixed before merging this one. |
|
Thanks! |
This is draft because I don't necessarily know that this is what we want to do yet.
See the discussion on issue #1119
When this is paired with a PR godotengine/godot#77410 to Godot that implements "possible solution nr 2", then both virtual and non-virtual methods are called with a consistent encoding (ie.
Object **) and seem to work in my limited testing :-)NOTE: This should fail tests until godotengine/godot#77410 is merged (if it is merged)