83517 - nsTransform2D's translation part can't be set in transformed coords

Status: RESOLVED FIXED

(Core Graveyard :: GFX, defect)

Description

[Alex Fritze]

There's an asymmetry in nsTransform2D's GetTranslation() and AddTranslation() 
functions, which makes it a pain to construct a transformation matrix that is 
equal to another matrix sans the translation part. (This is needed for the SVG, 
where we need to offset the rendering context's transformation to correctly map 
out a temporary offscreen surface.)

The problem is that nsTransform2D::GetTranslation() returns the translation 
part of the matrix in _already transformed_ units, i.e. it just returns the m20 
and m21 matrix elements.
nsTransform2D::AddTranslation()'s arguments, on the other hand, will be 
scaled/rotated by the matrix before they are added to m20/m21.
This means that, to get rid of the matrix translation offset while keeping the 
rest, you can't just do "AddTranslation( - GetTranslation() )".

It would help immensely if we could add a "SetTranslation()"-function that is 
the direct counterpart to GetTranslation() : 

Index: nsTransform2D.h
===================================================================
RCS file: /cvsroot/mozilla/gfx/public/nsTransform2D.h,v
retrieving revision 1.2
diff -u -r1.2 nsTransform2D.h
--- nsTransform2D.h	1999/11/06 03:20:49	1.2
+++ nsTransform2D.h	2001/05/31 19:25:11
@@ -105,6 +105,17 @@
   void GetTranslationCoord(nscoord *ptX, nscoord *ptY) { *ptX = NSToCoordRound
(m20); *ptY = NSToCoordRound(m21); }
 
  /**
+  * set the translation portion of this transform
+  *
+  * @param      tx, x translation
+  * @param      ty, y translation
+  * @exception
+  **/
+
+  void SetTranslation(float tX, float tY) { m20 = tX; m21 = tY; type |= 
MG_2DTRANSLATION; }
+
+  
+ /**
   * get the X translation portion of this transform
   *
   * @param

Comment 1

[Kevin McCluskey (gone)]

Reassigning to dcone.

Comment 2

[Bradley Baetz (:bbaetz)]

Attachment: patch with correct indenting

OK, this has been here for 5.5 months. I'm attaching a patch with the inlining
fixed.

Comment 3

[Bradley Baetz (:bbaetz)]

pav: can you review this?

Comment 4

[Stuart Parmenter]

Comment on attachment 57429 [details] [diff] [review]
patch with correct indenting

r=pavlov

Comment 5

[Johnny Stenback (:jst)]

Comment on attachment 57429 [details] [diff] [review]
patch with correct indenting

sr=jst with the style cleanup I talked to bbaetz about online.

Comment 6

[Bradley Baetz (:bbaetz)]

Checked in for alex, with a sensible style (ie not what the rest of the file does)