Done with the word "fix" for anymore detail - either adjust the commit message on your own. I am 100% bored of revisiting patches because someone wants to have a different commit message. Take it as it is here, or change it yourself, everything that i will write will look the same.
It says it fixes the up / down structure.
The patch adds a test + the fix. and that is expressed in the title and documentation. I do not see a big reason for splitting this.
I am fine with this.
Looks good otherwise :)
This seems again to explode on our CI. works with clang and gcc-8 not with gcc-9 ...
This here is a funny one, we are defining the event, but we are not emitting it at all, where or how should we emit this one ? In legacy it seems like we do not have something like this.
Does someone know where or how to emit this ?
okay - but mprotect has nothing to do as far as i can see with malloc_usable_size. So i would rather add there a guard for if mprotect is available or not (Or emulate mprotect with VirtualProtect?)
Thank you for pointing out how to search the repository, i still don't get what you are talking about.
That would just turn off more debugging that we can actually provide without malloc_usable_size, i don't think this is what we want ?
User error - sorry.
Mhm - the theme does not specify this property ... and as far as breakpoints tell, this is not set to false after this call...
Can you add the same for the unified widget ?
I just wanted to ensure we have thought about button as part, so this is fine to me :)
So action items for this here are :
- Change of Event type & Removal of struct type
- Write Tests. / Verify event emission / Veryfy funcitonal icon / str parts.
Sat, Jun 22
The one side is the API, the other side is how we are *using* this API from the widget.
- Calendar does use efl_ui_format_cb internally, which means, setting the format_string to the object *does* changes the format_cb, which looks super weird to me. Additionally, the format_cb does not handle all the format symbols that others do.
- efl_ui_spin only works with format_cb, setting the string does not work.
- efl_ui_tags only works with format_cb, setting the string does not work.
Why do we need align_priority ?
The event looks a bit overengeneered to me. Why passing a struct if its just a enum in the struct ? I think passing Efl.Ui.Alert_Popup_Button as event type is totally enough.
I am wondering a little bit about align / popup_size.
Fri, Jun 21
@sharkcz The grand council has decided that we should support this case, can you check the revision above ?
@segfaultxavi on my dell laptop i can hit shift 5 times which activates the shift-lock. Differentence between those two: Make caps lock on, and press "-". Make shift lock on (or just hold it) and press "-".
I am not against your proposal, it makes sense. However, its quite abstract, and i do not see how it will work later on in terms of API.
this time - at least try to include the fix ...
Furthermore, I'm not sure that your "active_gesture_object" idea is actually feasible; the edje callback which emits the original signals is before any of the widget callbacks, so it cannot be intercepted. This means the button will be in the "pressed" state. So now the move events start to occur, and "active_gesture_object" is set. How does the button stop being in the "pressed" state?
Well, its kind of an policy we started to use for building efl with meson. Everything is a explicit option, nothing is decided implicit. :/
Mhm, is your arch capable of any of the optimizations that you can see in this file ?
If not, then i think there is no native-arch-optimization for you, and -Dnative-arch-optimization=false should be applied ...
Thu, Jun 20
The idea of mike a little bit more in detail (continuing brainstorming):
To me this sounds like we are only running in this issue when we are capturing gestures which are decided "to be gestures" after the inital down event.
So how about this, we add a property to the scene, something like "active_gesture_object" if this is set, evas will move the mouse "virtually" outside the object it is in right now, so the objects will stop beeing in "down" mode. By the time there is a active_gesture_object, only this object will get the move events.
The active_gesture_object for us would then be scroller / active_view (and others?)
Ah! those are actually mirrored properties, for convinience usage i guess, sometimes just &'ing values is better, sometimes a property is easier ... :)
processed means that someone did something with this event, and none else should deal with it again.
also rename the enum
This is already stable.
There is no has_pointer_caps in efl_input_event ... ?
doc cop does doc cop things
phab transmitted it twice ... yey