View previous topic :: View next topic |
Author |
Message |
OPENSTEP The One
Joined: May 25, 2003 Posts: 4752 Location: Santa Barbara, CA
|
Posted: Thu May 17, 2007 9:25 am Post subject: |
|
I'm still seeing some very slight off-by-one errors in some Writer scrolling with some horizontal lines of single pixel artifacts being left behind. It's difficult to get them to appear with normal scrolling using the scrollbar, but they seem to be reproducible when using the scrollwheel a lot (e.g. fast up & down motions with scrollwheel).
Verified artifacts don't exist in stock patch 3, not sure if it's a boundary not being included in clipping or perhaps some type of timer issue. I'll try to find a better test case when I get home tonight.
ed |
|
Back to top |
|
|
Lorinda Captain Mifune
Joined: Jun 20, 2006 Posts: 2051 Location: Midwest, USA
|
Posted: Thu May 17, 2007 10:00 am Post subject: |
|
I can verify seeing the same artifacts. As Ed said, they were not there with patch 3; I first saw them in patch 7, although I skipped patch 6 and maybe patch 5.
I wasn't using the scroll wheel, but I was scrolling with my trackpad. I'm also seeing white lines--presumably one pixel high--running through some of the text. I can provide a screen shot (of both the artifacts and white lines) if that's helpful.
Also, there is a real lag scrolling with the track pad; the screen doesn't scroll with the scrolling of the track pad; then it jumps suddenly to a new position. I've noticed this for several test patches. Everytiime I came to report it, there was a new patch, and I figured it was better to check the new patch before reporting it.
Lorinda
(using PPC) |
|
Back to top |
|
|
pluby The Architect
Joined: Jun 16, 2003 Posts: 11949
|
Posted: Thu May 17, 2007 10:31 am Post subject: |
|
I believe that these artifacts and the delay in scrollbar drawing is due to the frequency that I use to flush pixels to the screen. Through the official Patch 3, I flushed after every single user event was handled. However, through all of the test patches I flushed only after all pending user events were handled. Hence, if you flood the event queue with lots of mouse events, flushing keeps getting delayed until you stop pushing more events in the queue.
To fix this, I have restored the flushing frequency back to the Patch 3 rate (i.e. after each single user event). Does the following test patch fix the problem?:
PowerPC:
http://www.neooffice.org/test/NeoOffice-2.1-Patch-3-Test-8-PowerPC.dmg
Intel:
http://www.neooffice.org/test/NeoOffice-2.1-Patch-3-Test-8-Intel.dmg
Patrick |
|
Back to top |
|
|
mickeyinlalaland Oracle
Joined: Feb 18, 2007 Posts: 239 Location: LaLaLand
|
Posted: Fri May 18, 2007 2:48 pm Post subject: |
|
Quote: | OPENSTEP wrote:
I'm still seeing some very slight off-by-one errors in some Writer scrolling with some horizontal lines of single pixel artifacts being left behind. |
I can't comment on that - as I've been working with slide shows:
What I can comment on NeoOffice-2.1-Patch-3-Test-8 is with the slide shows from bug 1473; Présentation de l'option Spécialité.sxi: and bug 2257; stats overview.ppt
Intel machine
— using slide show from bug 1473; has picked up speed in loading file and in transitions of slides and animations.
—using slide show from bug 2257; loads file quickly; loads slides smoothly and transitions between slides nicely; background image seems to begin appearance on page immediately; rest of elements (text is set to load on click of mouse) very responsive.
PPC machine:
— using slide show from bug 1473; seems to have slowed down a bit in opening file; transitions are smooth; slide loads but animations take ~3 seconds to load/begin play
—using slide show from bug 2257; I've noticed that the background image still loads evenly and I would say well; the slow-down appears when going to NEXT slide; the program transistion seems to lag; a blank white page holds the frame for ~3 seconds before background image loads and rest of slide elements transition; (text is set to load on click of mouse); so after slide transition lag with white page the rest of the elements for that 1 slide load nicely and smoothly - load time of text similar to OOo 2.1 X11)
X11 - OOo not loaded; off
Quit Neo and loaded OOo 2.1 X11
—using slide show from bug 1473; when compared to Neo screen transitions in OOo are a bit choppy; still smooth and yet transition can hiccup/skip and image has progressed further along in transition-slightly, this happens very fast but is noticeable; animations take about ~1sec to start playing.
—using slide show from bug 2257: slide transition has white page more or less appear as a flash of white before background image fills in; white page appears for 1/10th of a second; the rest of the elements (text is set to load on click of mouse) on the 1 slide transition nicely and similar to NeoO |
|
Back to top |
|
|
pluby The Architect
Joined: Jun 16, 2003 Posts: 11949
|
Posted: Sat May 19, 2007 2:11 pm Post subject: |
|
OPENSTEP wrote: | I'm still seeing some very slight off-by-one errors in some Writer scrolling with some horizontal lines of single pixel artifacts being left behind. It's difficult to get them to appear with normal scrolling using the scrollbar, but they seem to be reproducible when using the scrollwheel a lot (e.g. fast up & down motions with scrollwheel). |
I traced down the cause of this and it was our old friend the Java "non-rectangular clipping" bug and fixed it in the following test patch. Check out my commit comment for VCLGraphics.java for more details for where I changed the code to allow use to the non-rectangular clip area and when I have to paint in each separate clip rectangle.
In addition, I made the following two changes since the last test patch:
1. Since we get so many "I changed the paper size and/or orientation in the File :: Printer Settings dialog and the settings don't stick" postings, I hacked the native Cocoa Page Setup dialog that is displayed and, on Tiger and Leopard, the paper size and orientation controls in the dialog will be hidden. It's not pretty, but the key point that paper size and orientation are not set in that dialog should be more obvious with this change. Note: there is no change on Panther as Panther doesn't make the individual controls accessible from Cocoa.
2. Printing on the latest Leopard Developer Seed now works and, after finding the cause of this bug, it is quite clear that it didn't print before because the Apple Java programmers added a very stupid bug in the latest Developer Seed. Hint to Apple: read the JavaDocs for the Java API method that you are "optimizing" to understand the range of allowable return values before you decide to treat negative values as an error.
Anyway, here's the test patch URLs:
PowerPC:
http://www.neooffice.org/test/NeoOffice-2.1-Patch-3-Test-8-PowerPC.dmg
Intel:
http://www.neooffice.org/test/NeoOffice-2.1-Patch-3-Test-8-Intel.dmg
Patrick |
|
Back to top |
|
|
OPENSTEP The One
Joined: May 25, 2003 Posts: 4752 Location: Santa Barbara, CA
|
Posted: Sat May 19, 2007 2:24 pm Post subject: |
|
Thank you for all the work. I didn't make headway yesterday and will check things out.
Awful the Apple devs are still screwing with Java on Leopard and in nasty ways.
ed |
|
Back to top |
|
|
Lorinda Captain Mifune
Joined: Jun 20, 2006 Posts: 2051 Location: Midwest, USA
|
Posted: Sun May 20, 2007 2:51 pm Post subject: |
|
I notice that you've posted a link to a test patch 8 twice; did you reissue test patch 8, or should the second one be test patch 9? Or am I missing something obvious? |
|
Back to top |
|
|
pluby The Architect
Joined: Jun 16, 2003 Posts: 11949
|
|
Back to top |
|
|
Lorinda Captain Mifune
Joined: Jun 20, 2006 Posts: 2051 Location: Midwest, USA
|
Posted: Sun May 20, 2007 6:36 pm Post subject: |
|
pluby wrote: | FYI, when I upload a new test patch, I make the older ones redirect to the newest so following my duplicate posting will get you test patch 9: |
That's interesting. I caught the duplication because when I clicked on the second posting and chose to download, I was told I already had a file named patch 8, did I want to overwrite. But do I understand correctly that it really would have been patch 9?
Lorinda |
|
Back to top |
|
|
mickeyinlalaland Oracle
Joined: Feb 18, 2007 Posts: 239 Location: LaLaLand
|
Posted: Sun May 20, 2007 6:48 pm Post subject: |
|
Results from test NeoOffice_2.1_patch-3_test-9:
PPC machine:
— using slide show from bug 1473; transitions are smooth (choppier in OOo 2.1 as noted earlier); slide and animations take ~1 second to load/begin play this is more similar to behavior in OOo 2.1
—using slide show from bug 2257; most notable change from previous comment "with the program transistion seeming to lag"; a blank white page now only holds the frame for ~1 second before background image loads and rest of slide elements transition; (text is set to load on click of mouse)
No significant change to report on Intel machine:
As for printer settings; on PPC & Intel using a writer document; when I selected printer settings I get a window which flashes the blocked settings for only an instant before disappearing. |
|
Back to top |
|
|
pluby The Architect
Joined: Jun 16, 2003 Posts: 11949
|
Posted: Sun May 20, 2007 9:26 pm Post subject: |
|
Lorinda wrote: | That's interesting. I caught the duplication because when I clicked on the second posting and chose to download, I was told I already had a file named patch 8, did I want to overwrite. But do I understand correctly that it really would have been patch 9? |
Yes, this is correct.
Patrick |
|
Back to top |
|
|
sardisson Town Crier
Joined: Feb 01, 2004 Posts: 4588
|
Posted: Sun May 20, 2007 10:40 pm Post subject: |
|
pluby wrote: | Lorinda wrote: | That's interesting. I caught the duplication because when I clicked on the second posting and chose to download, I was told I already had a file named patch 8, did I want to overwrite. But do I understand correctly that it really would have been patch 9? |
Yes, this is correct.
Patrick |
BTW, you can always tell what you really have by looking at the title of the mounted disk image; it will always contain the right name even if you got the new patch from the redirect from the old patch's link.
Smokey _________________ "[...] whether the duck drinks hot chocolate or coffee is irrelevant." -- ovvldc and sardisson in the NeoWiki |
|
Back to top |
|
|
OPENSTEP The One
Joined: May 25, 2003 Posts: 4752 Location: Santa Barbara, CA
|
Posted: Mon May 21, 2007 10:50 am Post subject: |
|
Update: I was unable to get the one pixel artifact either yesterday when playing on my Quad and checked again on the machine where I originally spotted it (Dual 1.8 G5) and the artifacts do not appear here either with test patch 9.
ed |
|
Back to top |
|
|
amayze The Merovingian
Joined: Oct 24, 2005 Posts: 561 Location: Edinburgh, Scotland
|
Posted: Tue May 22, 2007 6:02 am Post subject: |
|
pluby wrote: | In addition, I made the following two changes since the last test patch:
1. Since we get so many "I changed the paper size and/or orientation in the File :: Printer Settings dialog and the settings don't stick" postings, I hacked the native Cocoa Page Setup dialog that is displayed and, on Tiger and Leopard, the paper size and orientation controls in the dialog will be hidden. It's not pretty, but the key point that paper size and orientation are not set in that dialog should be more obvious with this change. Note: there is no change on Panther as Panther doesn't make the individual controls accessible from Cocoa.
Patrick |
Downloaded Intel version of Patch 9 and Printer Settings Dialogue now only has Scale option.
But now I see Bugzilla/Trinity filling up with "Printer Settings dialogue is blank" bugs! Is there scope for the controls to be replaced with a few words to explain where to change the settings.
A button to take you to Format -> Page would be nice, but since that dialogue is specific to the Section that the insertion point is in, it might not be the best idea!
My 2p!
Andy. |
|
Back to top |
|
|
pluby The Architect
Joined: Jun 16, 2003 Posts: 11949
|
Posted: Tue May 22, 2007 8:30 am Post subject: |
|
amayze wrote: | But now I see Bugzilla/Trinity filling up with "Printer Settings dialogue is blank" bugs! Is there scope for the controls to be replaced with a few words to explain where to change the settings. |
That is no different than today when people keep changing the paper size and orientation in that dialog and can't figure out why their changes have no affect. Hiding the controls says that there is nothing to change there.
amayze wrote: | A button to take you to Format -> Page would be nice, but since that dialogue is specific to the Section that the insertion point is in, it might not be the best idea! |
I was hoping to display the Format :: Page dialog but this is not feasible. There is no handle in vcl to this specific menu item and each application implements calling of the dialog differently. The only option is to guess which menu is the right one by its name and since that is translated into 70+ languages, even testing if that would work is a non-trivial amount of work.
Patrick |
|
Back to top |
|
|
|