Issue 75936 - rect with draw:transform="rotate(99) missing part of top bounding box
Summary: rect with draw:transform="rotate(99) missing part of top bounding box
Status: CLOSED DUPLICATE of issue 75890
Alias: None
Product: Draw
Classification: Application
Component: viewing (show other issues)
Version: OOo 2.2
Hardware: All Windows XP
: P3 Trivial (vote)
Target Milestone: OOo 3.0
Assignee: Armin Le Grand
QA Contact: issues@graphics
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-03-30 21:36 UTC by asiansteev
Modified: 2007-04-11 16:30 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
Screenshot of rotated rectangle with missing part (5.56 KB, text/plain)
2007-03-30 21:37 UTC, asiansteev
no flags Details
Condensed version of test, with only broken rectangle (7.50 KB, text/plain)
2007-03-30 21:38 UTC, asiansteev
no flags Details
Full test with all rotations (13.17 KB, text/plain)
2007-03-30 21:39 UTC, asiansteev
no flags Details
New screen/diagram (51.18 KB, text/plain)
2007-03-30 21:52 UTC, asiansteev
no flags Details
screenshot of bug sequence with new testcase (simpler_demonstration.odg) (43.87 KB, text/plain)
2007-04-06 06:22 UTC, asiansteev
no flags Details
a simpler testcase that should demonstrate this issue (no rotations) (6.55 KB, text/plain)
2007-04-06 06:23 UTC, asiansteev
no flags Details

Note You need to log in before you can comment on or make changes to this issue.
Description asiansteev 2007-03-30 21:36:31 UTC
In the attached test, above the standard viewable page (scroll up), there is a
rotated rectangle.  Part of the bounding box is not being drawn, as can be seen
in the attached screen shot.  I have tried this on two seperate machines.  Do
not "page up" to bring the rectangle into view, as this will force a redraw. 
Instead, arrow up, or scroll with the arrow on the top of the right side scroll bar.
I have attached the full rotate test with many angles, and a condensed version
which has only the broken transform, draw:transform="rotate (99)"
Thanks!
Comment 1 asiansteev 2007-03-30 21:37:23 UTC
Created attachment 44082 [details]
Screenshot of rotated rectangle with missing part
Comment 2 asiansteev 2007-03-30 21:38:36 UTC
Created attachment 44083 [details]
Condensed version of test, with only broken rectangle
Comment 3 asiansteev 2007-03-30 21:39:53 UTC
Created attachment 44084 [details]
Full test with all rotations
Comment 4 asiansteev 2007-03-30 21:50:58 UTC
I notice now that when scrolling up, the line that disappears barely grazes the
top of the window, and this seems to cause the disappearing.  This also happens
on  74.  See the new attached screen with diagram
Comment 5 asiansteev 2007-03-30 21:52:48 UTC
Created attachment 44085 [details]
New screen/diagram
Comment 6 wolframgarten 2007-04-05 11:57:51 UTC
Sorry, not reproducible here. Please make sure that you have the latest graphic
card drivers installed.
Comment 7 asiansteev 2007-04-06 06:20:53 UTC
I'm attaching a newer file that I have verefied on all 3 PCs that I have access
to, each with the latest video card driver installed (windows XP).  With OOo2.2,
the rectangles on this file should reach the top of the viewable drawing area. 
Just press the 'up' arrow button, and the tops of the rects should disappear. 
I'm sure this is resolution dependent, but it should not be that hard to reproduce.
Hope this helps.  Let me know.
Comment 8 asiansteev 2007-04-06 06:22:45 UTC
Created attachment 44243 [details]
screenshot of bug sequence with new testcase (simpler_demonstration.odg)
Comment 9 asiansteev 2007-04-06 06:23:33 UTC
Created attachment 44244 [details]
a simpler testcase that should demonstrate this issue (no rotations)
Comment 10 wolframgarten 2007-04-11 09:16:58 UTC
Thanks for the new file, this helped! Reproducible now.
What I found: opening the file with OOo2.1 shows the entire drawing object.
Opening the file with OOo2.2 the drawing object is directly placed at the upper
drawing area border. But in both cases the same format, zoomfactor and window
size is used. I do not know where there is the difference.
Reassigned.
Comment 11 Armin Le Grand 2007-04-11 16:29:55 UTC
AW: It has nothing to do with rotations of 99.
BTW: the draw:transform="rotate (XXX)" value is radians, so it will just take
99%(2*PI) as rotation anyways.
It's just another repaint inconvenience (use CTRL+ALT+R to see that it's not a
file format/model problem)  and thus very similar to #i75890#. We simply have
the problem that the conversions between logic and pixel coordinates CANNOT be
numerically correct since both are signed integers. Those annoyances are reduced
as much as possible, but until we have a better geometry base, there is no
solution and there will always be litte repaint annoyances. All in all te screen
isplay is a discrete visualisation of vetor data.
AW: Setting to duplicate to #i75890#...

*** This issue has been marked as a duplicate of 75890 ***
Comment 12 Armin Le Grand 2007-04-11 16:30:30 UTC
AW: Closing...