fix work with TRACK_POINTERS enabled #358
No reviewers
Labels
No labels
Accessibility
Accessibility
Address
Address
Android
Android
Android Auto
Android Auto
Android Automotive (AAOS)
Android Automotive (AAOS)
API
API
AppGallery
AppGallery
AppStore
AppStore
Battery and Performance
Battery and Performance
Blocker
Blocker
Bookmarks and Tracks
Bookmarks and Tracks
Borders
Borders
Bug
Bug
Build
Build
CarPlay
CarPlay
Classificator
Classificator
Community
Community
Core
Core
CrashReports
CrashReports
Cycling
Cycling
Desktop
Desktop
DevEx
DevEx
DevOps
DevOps
dev_sandbox
dev_sandbox
Directions
Directions
Documentation
Documentation
Downloader
Downloader
Drape
Drape
Driving
Driving
Duplicate
Duplicate
Editor
Editor
Elevation
Elevation
Enhancement
Enhancement
Epic
Epic
External Map Datasets
External Map Datasets
F-Droid
F-Droid
Fonts
Fonts
Frequently User Reported
Frequently User Reported
Fund
Fund
Generator
Generator
Good first issue
Good first issue
Google Play
Google Play
GPS
GPS
GSoC
GSoC
iCloud
iCloud
Icons
Icons
iOS
iOS
Legal
Legal
Linux Desktop
Linux Desktop
Linux packaging
Linux packaging
Linux Phone
Linux Phone
Mac OS
Mac OS
Map Data
Map Data
Metro
Metro
Navigation
Navigation
Need Feedback
Need Feedback
Night Mode
Night Mode
NLnet 2024-06-281
NLnet 2024-06-281
No Feature Parity
No Feature Parity
Opening Hours
Opening Hours
Outdoors
Outdoors
POI Info
POI Info
Privacy
Privacy
Public Transport
Public Transport
Raw Idea
Raw Idea
Refactoring
Refactoring
Regional
Regional
Regression
Regression
Releases
Releases
RoboTest
RoboTest
Route Planning
Route Planning
Routing
Routing
Ruler
Ruler
Search
Search
Security
Security
Styles
Styles
Tests
Tests
Track Recording
Track Recording
Translations
Translations
TTS
TTS
UI
UI
UX
UX
Walk Navigation
Walk Navigation
Watches
Watches
Web
Web
Wikipedia
Wikipedia
Windows
Windows
Won't fix
Won't fix
World Map
World Map
No milestone
No project
No assignees
1 participant
Due date
No due date set.
Dependencies
No dependencies set.
Reference: organicmaps/organicmaps-tmp#358
Loading…
Add table
Reference in a new issue
No description provided.
Delete branch "drape-ptr-refs"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
fixes #357
@ -74,2 +74,3 @@
void Remove(ref_ptr<OverlayHandle> handle);
//! \return true if tree completely invalidated and next call has no sense
bool Remove(ref_ptr<OverlayHandle> handle);
void EndOverlayPlacing();
Комментарий, что возвращается, не помешал бы.
@ -35,2 +35,3 @@
void RemoveOverlayHandles(ref_ptr<OverlayTree> tree);
//! \return true if tree completely invalidated and next call has no sense
bool RemoveOverlayHandles(ref_ptr<OverlayTree> tree);
void SetOverlayVisibility(bool isVisible);
И здесь.
static_assert?
@ -35,2 +35,3 @@
void RemoveOverlayHandles(ref_ptr<OverlayTree> tree);
//! \return true if tree completely invalidated and next call has no sense
bool RemoveOverlayHandles(ref_ptr<OverlayTree> tree);
void SetOverlayVisibility(bool isVisible);
Done
@ -74,2 +74,3 @@
void Remove(ref_ptr<OverlayHandle> handle);
//! \return true if tree completely invalidated and next call has no sense
bool Remove(ref_ptr<OverlayHandle> handle);
void EndOverlayPlacing();
Done
depthLayer не constexpr, поэтому
assert(depthLayer < kStateExtensions.size())
не может быть вычислен в compile-timeВерно, я сообразил и вроде удалял коммент, но гитхаб глюкнул.
Все еще иногда сообщает об удалении
unique_ptr
в то время как есть соответсвующиеref_ptr
разделяющие с ним указатель. Это происходит при остановке приложения, когда вызываетсяBackendRenderer::OnContextDestroy
.Так в очереди на обработку еще могут лежать
MapShape
у которых есть ссылка на данныеTextureManager
,который как раз очищается в
BackendRenderer::OnContextDestroy
. Но это вроде бы не критично, так какBackendRenderer::OnContextDestroy
вызывается только при завершении приложения и перед удалением чего-нибудьтам включается фильтр для всего связанного с графическим контекстом.
@ -160,1 +161,4 @@
Clear();
return true;
}
Hm, Is it a valid state or something went wrong and we made this patch? When
@ -160,1 +161,4 @@
Clear();
return true;
}
Yes, there are several places where
m_frameCounter
setted tokInvalidFrame
without call ofClear()
, for exampleOverlayTree::SetVisualScale
.And this is ok, because of only
OverlayTree::Remove
call can cause danglingref_ptr
.So with something like map scale change and move we can get kInvalidFrame via
OverlayTree::SetVisualScale
and only then call toOverlayTree::Remove
before start of next frame. And in that case we get dangingref_ptr
without extraClear
call.