Opened 13 years ago
Last modified 13 years ago
#3696 new bug
classification keeps remembering a removed feature class
Reported by: | rduivenvoorde | Owned by: | nobody |
---|---|---|---|
Priority: | major: does not work as expected | Milestone: | Version 1.7.0 |
Component: | Symbology | Version: | 1.6.0 |
Keywords: | Cc: | ||
Must Fix for Release: | No | Platform: | All |
Platform Version: | Awaiting user input: | no |
Description
when using unique classification, the removal of features from a certain class makes this class NOT go away when you re-classify the layer.
Reproduction:
- open qgis and generate a fresh point-shapefile with one string attribute 'foo'
- in editing mode digitize 3 features all with another value for foo (aaa, bbb, ccc)
- now apply unique classification, you should see three classes
- now remove one of the features using the edit/delete feature functionality in attribute-table view (ending up with only two features: aaa, bbb)
- now reclassify the layer: ccc keeps coming up as a class/value
- I have seen an example in which even restarting qgis this kept being the case (?????)
This happens in both 'new' and 'old' classification
I would see this as a bug, people using this classification as an legend in a layout will have non valid values in there legends
Or is there another explanation? (my explanation? somewhere there is a caching mechanism which is not refreshed? (let's forget the restarting of qgis... don't know what that was....).
Tested with 1.6 on windows and in debian
Change History (3)
comment:1 by , 13 years ago
follow-up: 3 comment:2 by , 13 years ago
running r15651 trunk here on ubuntu: still this problem
One thing: after saving the shapefile I also saved the project. (in real it was actually an excisting project).
can it be something with handles to the projectfile (which also holds the classification?)
comment:3 by , 13 years ago
Replying to rduivenvoorde:
running r15651 trunk here on ubuntu: still this problem
One thing: after saving the shapefile I also saved the project. (in real it was actually an excisting project).
can it be something with handles to the projectfile (which also holds the classification?)
reply to self:
sorry marco did not read your reply good enough. Indeed in trunk NEW symbology seems to work. OLD symbology not working.
IF this is the last 1.* version we can decide to leave it as is.... and spend energy on 2.* bugs ...
From a quick test with new symbology, this seems to work ok with current trunk.