Improve mitk::WeakPointer
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	kislinsk
	Nov 30 2017, 8:06 PM

Description

mitk::WeakPointer fills in the blank of a missing implementation of a true weak pointer for itk::Object. While there is an itk::WeakPointer, it is simply a raw pointer wrapper and of questionable use. The current implementation of mitk::WeakPointer still has some flaws, though:

It has more than a single responsibility by providing mitk::Messages for itk::ModifiedEvent and itk::DeleteEvent of the wrapped itk::Object.
It is not as efficient as it should be.
It can't be used in a natural way like a pointer because it lacks many comparison operators.
It provides direct and dangerous access to its encapsulated resource, the wrapped raw pointer.

I want to replace the current implementation of mitk::WeakPointer with a new, well thought out, high-quality implementation.

Related Objects
Search...

Status	Assigned	Task
Open	None	T27807 [MultiLabel Segmentation] Inspection of module and plugin
Open	None	T22593 Update the MitkMultilabelTestDriver
Open	None	T22458 Merging a label in the MultiLabel segmentation plugin leaves one label without a segmentation
Open	None	T26397 Multilabel Segmentation file cannot be converted to multilabel segmentation in MITK
Resolved	floca	T23981 Multilabel color map should disable interpolation
Open	None	T25445 MultiLabel segmentation mapper is very slow
Resolved	kalali	T28131 [MultiLabel Segmentation] 3D Threshold tool does not work with multiple labels / layers
Open	None	T28176 [Segmentation] Inspection of module and plugin
Resolved	floca	T28526 [Segmentation] Check tools (especially 3D tools) to correctly handle updating of LabelSetImage
Open	None	T26959 [Segmentation] Newly created LabelSet segmentation is not displayed as "outline binary"
Open	None	T28479 [Segmentation] Rethink / reevaluate our existing segmentation tools / algorithms
Open	None	T8140 Reimplement volume annotations for segmentations
Resolved	kislinsk	T27319 [Segmentation] Correctly implement interpolation features for Multilabel Segmentation View
Open	floca	T28524 Refactor LabelSetImage and associated classes.
Resolved	floca	T23721 Project "Astonishing Angelfish"
Resolved	floca	T23742 Implement proof of concept for new segmentation data type
Resolved	kislinsk	T23867 Improve mitk::WeakPointer

Event Timeline

kislinsk triaged this task as Wishlist priority.Nov 30 2017, 8:06 PM

kislinsk created this task.

kislinsk added a parent task: T23742: Implement proof of concept for new segmentation data type.

Pushed new branch T23867-WeakPointer.

Migration Guide

The public API of mitk::WeakPointer<T> changed completely. It follows the idea of std::weak_ptr<T> now. Access to the object is gained by converting it to T::Pointer resp. itk::SmartPointer<T>. To make clear that the whole principle of mitk::WeakPointer<T> changed, the wording of the public API is also based on std::weak_ptr<T> now.

Check if the weak pointer is expired

Old API

if (weakPointer.IsNull())
  ...

New API

if (weakPointer.IsExpired())
  ...

// or

if (!weakPointer)
  ...

// or

if (nullptr == weakPointer)
  ...

Access the object

Old API

if (weakPointer.IsNotNull())
  weakPointer->Foo();

New API

if (weakPointer)
{
  auto object = weakPointer.Lock();
  object->Foo();
}

Observe events

You can get register a single callable for the deletie event. You can use another weak pointer if you need to have more than one hook or observe the wrapped object directly instead of using mitk::WeakPointer<T> as a proxy.

kislinsk updated the task description. (Show Details)Nov 30 2017, 11:39 PM

kislinsk mentioned this in rMITK76463123c0e1: Merge branch 'T23867-WeakPointer'.Dec 1 2017, 1:24 AM

kislinsk mentioned this in rMITK63c57e902c93: Prepare merge of T23867.Dec 1 2017, 1:26 AM

kislinsk mentioned this in rMITKfc719a8c0a45: Merge branch 'T23867-WeakPointer' into T23742-NewSegmentationDataType.

kislinsk closed this task as Resolved.Dec 1 2017, 1:26 AM

kislinsk mentioned this in rMITKea4e606c0eec: Merge branch 'T23867-WeakPointer'.Dec 1 2017, 1:42 AM

kislinsk mentioned this in rMITK501d6d07b244: Merge branch 'T23867-WeakPointer' into T23742-NewSegmentationDataType.

kislinsk mentioned this in rMITK9e6dd9c88011: Merge branch 'T23867-WeakPointer'.Dec 4 2017, 4:57 PM

kislinsk mentioned this in rMITKa4444d6e78a8: Merge remote-tracking branch 'origin/T23867-WeakPointer' into T23742….Dec 4 2017, 7:28 PM

kislinsk mentioned this in rMITK1e2caa2db74b: Merge branch 'T23867-WeakPointer'.Dec 4 2017, 10:39 PM

kislinsk mentioned this in rMITK57c4b9327548: Merge remote-tracking branch 'origin/T23867-WeakPointer' into T23742….Dec 4 2017, 10:45 PM