Version: 1.1.1 (using KDE KDE 3.1) Installed from: Mandrake RPMs OS: Linux Association anchor point deletion is buggy. 1) Wrong anchor point deleted * Make two classes * Draw an association between them * Put two anchor points in the association arrow between the two association endpoints * Double-click on the new anchor point farthest from the arrow head. The other anchor point will be deleted. It seems that it is always the anchor point next closest the arror head that is deleted when you double-click on an anchor point 2) Self-associations-- can delete all interior anchor points * Make class * Make (unidirectional) self-association to that class, with rectangular arrow with two non-endpoint anchor points. You can delete both of these non-endpoint anchor points, resulting in a very odd arrow. Any labels on the arrow are left dangling. 3) Anchor point boxes don't go away * In many cases, just experimenting as above, you'll find that the blue anchor point boxes don't go away when the anchor is deleted.
The problem is the wrong understanding of m_LineList and the return value of onLinePath, both in class LinePath. removePoint always removes the start point of the line, but this might not be always the correct behaviour. I'll fix this in the next 2 weeks. Steinchen
Subject: Re: association anchor deletion buggy the following 2 items are now fixed in CVS: > 1) Wrong anchor point deleted > 3) Anchor point boxes don't go away I won't fix the following part of the bug report: > 2) Self-associations-- can delete all interior anchor > points > * Make class > * Make (unidirectional) self-association to that > class, with rectangular arrow with two non-endpoint anchor points. > You can delete both of these non-endpoint anchor points, resulting in > a very odd arrow. Any labels on the arrow are left dangling. Because if the user really wants to do so, his problem. Even if we restrict the number of points of self-associations to be at least 3, he can always move the middle point on a line between the outer points. This would be the same like deleting the middle point.
*** Bug has been marked as fixed ***.
As you say, 1 and 3 seem much better in v. 1.2.1. I'm not going to press point 2 -- it's a judgement call.