After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 303301 - Center point far too magnetic
Center point far too magnetic
Status: RESOLVED FIXED
Product: dia
Classification: Other
Component: general
CVS head
Other All
: High major
: 0.95
Assigned To: Dia maintainers
Dia maintainers
: 307519 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2005-05-06 20:49 UTC by Hans Breuer
Modified: 2005-10-15 14:24 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Hans Breuer 2005-05-06 20:49:13 UTC
Again try testing with an UML diagram. Connecting
a class with an association is almost impossible
at the moment. Beside the broken autorouting there
seems to be a problem with the new center points.

Or maybe some connection points are got lost
during conversion (mid top and mid bottom)?
Comment 1 Hans Breuer 2005-05-08 21:59:30 UTC
After some hours of debugging, result applied to cvs :
[Part of the general problem is distance_rectangle_point()
used in most DiaObject::distance_from() but returning 0.0 
for every point with-in the rectangle. So while trying to
connect an 'border' connection point don't get into the 
bounding box ...]

2005-05-08  Hans Breuer  <hans@breuer.org>

	* objects/UML/class.c : Reverted to have no center point.
	At first it looked as if there was only the bug to not increase
	class.h:UMLCLASS_CONNECTIONPOINTS. But the UMLClass has 
	potentially much more dynamic connection points (to the left
	and right of every attribute/operation). Now simply adding one
	between see static and dynamic will screw up any existing diagram
	which makes use of the dynamic connection points. The would all
	be connected with an off-by-one error. I have no idea how to
	code around that. But breaking backward compatibility is no 
	option here. Does not completely fix bug #303301 ...
	
Comment 2 Lars Clausen 2005-05-09 19:45:52 UTC
Oh, dear.  That probably means the main CP will have to be after the dynamic
ones.  Bugger.
Comment 3 Lars Clausen 2005-06-02 20:21:02 UTC
Should be fixed with current CVS (when I can get the commit through!).  Leaving
the #ifdef in until release time in case more problems come up, but testing it
gave no problems.
Comment 4 Hans Breuer 2005-06-10 10:32:40 UTC
Hi Lars,
Probably I should have been more explicit with my fix - or to 
rephrase it - this should have been *two* bug reports.

1) Centerpoint far too magnetic (not UML Class specific!)
2) Midpoint break UMLs Class compatibility

While it seems your patch addresses the latter the former 
behaviour is not changed at all.

2005-06-02  Lars Clausen  <lars@raeder.dk>

	* objects/UML/class_dialog.c:
	* objects/UML/class.[ch]:
	The final fixes for mainpoint on UML Class.


Please try to change the connections in samples/UMLPackages.dia or
try to reconnect them. Reopening ...
Comment 5 Hans Breuer 2005-07-10 16:55:49 UTC
2005-07-10  Hans Breuer  <hans@breuer.org>

	* app/connectionpoint_ops.c app/object_ops.c : make the
	'whole object' connection point only center point again. 
	That is: it is visible and far less magnetic. Only the
	line gap handling make it special. For me this probably
	fixes bug #303301

Comment 6 Hans Breuer 2005-08-05 16:16:06 UTC
*** Bug 307519 has been marked as a duplicate of this bug. ***
Comment 7 Hans Breuer 2005-08-05 16:17:06 UTC
The dup has a nice use case I reused here:
http://mail.gnome.org/archives/dia-list/2005-August/msg00024.html
Comment 8 Lars Clausen 2005-09-29 05:41:47 UTC
The example shown in the dup is not dependent on mainpoints, it happens whether
or not they are turned on.  Thus, I am enabling them, given that the other
concerns in here have been addressed.  If there are other problems with main
points, please add more bugreports, this one is now officially about problems
with sattributes/operations on UML object having connection issues.
Comment 9 Hans Breuer 2005-10-15 14:24:13 UTC
Lars: I dont agree with your attempt to redefine this bug. Closing 
it cause in fact the main issue with too magnetic center points 
is fixed. If you care enough un-dup the other bug.