Bug 75065

Summary:	Transformed symbols sometimes disappear
Product:	ksvg	Reporter:	Jeff Smith <whydoubt>
Component:	general	Assignee:	Nikolas Zimmermann <wildfox>
Status:	RESOLVED WORKSFORME
Severity:	normal
Priority:	NOR
Version:	unspecified
Target Milestone:	---
Platform:	Compiled Sources
OS:	Linux
Latest Commit:		Version Fixed In:
Attachments:	Testcase should show four circles Patches bug in transforming symbols

Description Jeff Smith 2004-02-12 13:49:36 UTC

Version:            (using KDE KDE 3.2.0)
Installed from:    Compiled From Sources
Compiler:          gcc 3.2 
OS:          Linux

Certain transformations in a 'use' cause the entire symbol not to be drawn.  For instance translate(-1,1) or translate(1,-1).

After a lot of digging in code I found the cause: the clipping box is being flipped by the transform.  This causes everything *inside* the box to be clipped when it should be clipping everything *outside* the box.

I have a patch which fixes this by checking if the box is clockwise after the affine transformation.  If clockwise, flipping two of the vertices corrects it.

Comment 1 Jeff Smith 2004-02-12 14:01:54 UTC

Created attachment 4651 [details]
Testcase should show four circles

Comment 2 Jeff Smith 2004-02-12 14:06:14 UTC

Created attachment 4652 [details]
Patches bug in transforming symbols

Comment 3 Nikolas Zimmermann 2004-02-12 16:28:43 UTC

On Thursday 12 February 2004 13:49, Jeff Smith wrote:
>
> Certain transformations in a 'use' cause the entire symbol not to be drawn.
>  For instance translate(-1,1) or translate(1,-1).
>
> After a lot of digging in code I found the cause: the clipping box is being
> flipped by the transform.  This causes everything *inside* the box to be
> clipped when it should be clipping everything *outside* the box.
>
> I have a patch which fixes this by checking if the box is clockwise after
> the affine transformation.  If clockwise, flipping two of the vertices
> corrects it.

Oh my god :) Thank you very much! Already thought about active ksvg
development? I'd appreciate it...

Bye
 Bye
  Niko


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFAK5rzs77DQgSFsQERAsKgAJ4oVsZ4HzcLgrKJQZ+SzwjQOXP0eACfdXRb
zjPcYbuc9A2fNwBMo0g7e80=
=cEI5
-----END PGP SIGNATURE-----

Comment 4 Jeff Smith 2004-02-19 05:17:12 UTC

I fixed this in kdegraphics/ksvg/plugin/backends/libart/LibartCanvas.cpp.
I found that I could do this based on the sign of the determinant of the
transformation matrix, rather than using trig on the vector product.