Marble/GeoData/PointerVsImplicitShare: Difference between revisions
(Started this page ... first thoughts, multiple possibilities) |
m (6 revisions imported: Import from techbase...) |
||
(5 intermediate revisions by 2 users not shown) | |||
Line 6: | Line 6: | ||
To find out which is the wisest I try to look at every single class: | To find out which is the wisest I try to look at every single class: | ||
(please add your thoughts here) | |||
=== Base data === | |||
* '''GeoDataCoordinates''' This one is quite simple. It is not more than a QPoint in GeoCoordinates. This should stay implicilty shared. | |||
* '''GeoDataLatLonBox''' (GeoDataObject) It seems to nothing more than a QRect for GeoCoordinates. I'm not sure if it must be derived from GeoDataObject | |||
** '''GeoDataLatLonAltBox''' same for this | |||
* '''GeoDataAccuracy''' Should be base data => value | |||
* '''GeoDataLod''' | |||
* '''GeoDataRegion''' I'm not sure if it belongs into this group ... | |||
=== Geometric structures === | |||
* '''GeoDataGeometry''' The base class for all geometric. We could want a pointer for this, so using it as data may not be the right way. | |||
** '''GeoDataPoint''' This one uses multiple inheritance (GeoDataCoordinates). I don't think that this is semantically correct. | |||
** '''GeoDataLineString''' | |||
*** '''GeoDataLinearRing''' | |||
** '''GeoDataMultiGeometry''' | |||
** '''GeoDataPolygon''' | |||
=== Features === | |||
* '''GeoDataFeature''' These should be pointered, so GeoGraphicsItems could refer to them and change them easily. | |||
** '''GeoDataFeature''' | |||
** '''GeoDataContainer''' | |||
*** '''GeoDataFolder''' (by the way, this is an object which probably wouldn't need a GeoGraphics representation) | |||
*** '''GeoDataDocument''' (GeoDocument) | |||
=== views === | |||
* '''GeoDataAbstractView''' It's nothing more than a view, this could stay implicitly shared. | |||
** '''GeoDataLookAt''' Same for this as this is a child. | |||
=== styles === | |||
* '''GeoDataColorStyle''' | |||
** '''GeoDataIconStyle''' | |||
** '''GeoDataLabelStyle''' | |||
** '''GeoDataLineStyle''' | |||
** '''GeoDataPolyStyle''' | |||
* '''GeoDataStyleSelector''' | |||
** '''GeoDataStyle''' | |||
** '''GeoDataStyleMap''' (QMap<QString, QString>) | |||
* '''GeoDataHotSpot''' | |||
It is possible to group these things in two groups: | |||
* objects that represent data which is actually shown on the map: | |||
** Features | |||
* objects that are abstract | |||
** everything else :) | |||
I'm not sure to which of the groups the geometry classes belongs. | |||
A compromise could be to use the classes of the first group with a pointer and all others as a value. | |||
In KML all items are children of GeoDataObject (except GeoDataCoordinates). See [http://code.google.com/intl/de-DE/apis/kml/documentation/kmlreference.html#object here]. |
Latest revision as of 16:40, 25 October 2016
Implicit sharing and storing objects as value is very reasonable for basic data as QStrings, but there are reasonable doubts regarding Placemarks and other GeoData. There are some possibilities:
- Don't use implicit sharing at all and store all objects on the heap with a pointer to it.
- Use implicit sharing and storing as value for GeoDataCoordinates only.
- Use implicit sharing and storing as value for all GeoData classes.
To find out which is the wisest I try to look at every single class: (please add your thoughts here)
Base data
- GeoDataCoordinates This one is quite simple. It is not more than a QPoint in GeoCoordinates. This should stay implicilty shared.
- GeoDataLatLonBox (GeoDataObject) It seems to nothing more than a QRect for GeoCoordinates. I'm not sure if it must be derived from GeoDataObject
- GeoDataLatLonAltBox same for this
- GeoDataAccuracy Should be base data => value
- GeoDataLod
- GeoDataRegion I'm not sure if it belongs into this group ...
Geometric structures
- GeoDataGeometry The base class for all geometric. We could want a pointer for this, so using it as data may not be the right way.
- GeoDataPoint This one uses multiple inheritance (GeoDataCoordinates). I don't think that this is semantically correct.
- GeoDataLineString
- GeoDataLinearRing
- GeoDataMultiGeometry
- GeoDataPolygon
Features
- GeoDataFeature These should be pointered, so GeoGraphicsItems could refer to them and change them easily.
- GeoDataFeature
- GeoDataContainer
- GeoDataFolder (by the way, this is an object which probably wouldn't need a GeoGraphics representation)
- GeoDataDocument (GeoDocument)
views
- GeoDataAbstractView It's nothing more than a view, this could stay implicitly shared.
- GeoDataLookAt Same for this as this is a child.
styles
- GeoDataColorStyle
- GeoDataIconStyle
- GeoDataLabelStyle
- GeoDataLineStyle
- GeoDataPolyStyle
- GeoDataStyleSelector
- GeoDataStyle
- GeoDataStyleMap (QMap<QString, QString>)
- GeoDataHotSpot
It is possible to group these things in two groups:
- objects that represent data which is actually shown on the map:
- Features
- objects that are abstract
- everything else :)
I'm not sure to which of the groups the geometry classes belongs.
A compromise could be to use the classes of the first group with a pointer and all others as a value. In KML all items are children of GeoDataObject (except GeoDataCoordinates). See here.