bugXnee - Bugs: bug #32385, CNEE replay precision

 
 

bug #32385: CNEE replay precision

Submitted by:  Kiran D <kiran_kdmukh>
Submitted on:  Tue 08 Feb 2011 12:45:43 PM UTC  
 
Category: NoneSeverity: 4 - Important
Item Group: NoneStatus: Fixed
Privacy: PublicAssigned to: Henrik Sandklef <hesa>
Open/Closed: Closed

Add a New Comment(Rich Markup)
   

You are not logged in

Please log in, so followups can be emailed to you.

 

(Jump to the original submission Jump to the original submission)

Thu 12 May 2011 10:37:19 AM UTC, comment #10:

When comparing images* I am now down to 8 when doing

compare -metric MAE ~/xnee-xorg.png xnee-xorg.png diff.png

The images are much more detailed than the one used by Kiran. Comparing Kiran's images gives 1.75 ... but are at the same time way less detailed.

Getting closer ....

*) as drawn with mtpaint while both recording and replaying

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Wed 11 May 2011 10:13:14 PM UTC, comment #9:

Xnee needs to take the following into consideration:

1. Time since start
2. Time between events
3. Synching with applications (e.g wait for new window, wait for data as received over network)

Also a normal Desktop OS does not provide an "accurate implementation of usleep".

I have rewritten (and fixed a bug in) the function that calculates the delay so that I now get "only" get a distortion of 22 with compare*. This and the rather slight diff when comparing the times since start (recording and replayin) of 269 usecs of a 20 seconds session.

I will do some (read many many more) tests and get back. Just wanted to give the feedback I have.

Nothing is committed to CVS.

/h

*) compare -metric MAE ~/xnee-xorg.png xnee-xorg.png diff.png

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Tue 10 May 2011 07:53:44 PM UTC, comment #8:

I record an mtpaint session.
Replay it and record it (again) while replaying.

Comparing the delays between the fakes is somewhat strange ...

  • there is a diff in time since start

this should NOT happen

  • there is a bit difference between the delays between the individual event fakes...but not alarming. But this diff is most likely adding up to the diff above

Strange says indeed!

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Tue 10 May 2011 06:15:12 PM UTC, comment #7:

Work begun in branch:

savannah_32385

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Sun 27 Mar 2011 08:57:27 PM UTC, comment #6:

I can now reproduce the error, using mtpaint, import and compare.

/h

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Sat 26 Mar 2011 09:04:24 PM UTC, comment #5:

This is a bit confusing ... perhaps it is a timing issue.
I will check in to it more ..

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Fri 25 Feb 2011 11:54:02 AM UTC, comment #4:

I installed SUSE Linux 11.3(SLED) and installed the Latest XNEE available Xnee-3.09.
The problem still exists.
Here I used the "mtPaint Image Editor" which comes installed with OS installation.

Here is the steps to be followed to reproduce this. Sorry if it is more descriptive than required.

Steps for reproducing this:-

1. Start a cnee record session "cnee --record --keyboard --mouse --out-file record1.xnl --events-to-record -1 --stop-key q"
2. Press "Computer" Button in the left bottom corner of screen
3. Press "More Applications..." and Browse through the opened Window.
4. Select "mtPaint Image Editor" application under Graphics.
5. Click on the "Paint" tool (looks like a pencil) and draw few curves on the image area.
6. Take a Screenshot using PrintScreen key from the keyboard.
7. A small window will pop up to confirm save. Save it in the desired location.
8. Now press q and stop the record session.
9. Move the Saved image from the current location to another location.
10. Start the replay session "cnee --replay --file record1.xnl"
11. Image is saved again in the same location.
12. Compare both the Screenshots taken at the time of record and replay.

Here i use a home made Pixel by Pixel comparison code from java, which also produces a difference image.

I would also suggest you to use Paint(OpenOffice.org) in the similar way.

I have attached 3 images which depict the scenario.

Thanks

(file #22780, file #22781, file #22782)

Kiran D <kiran_kdmukh>
Sat 19 Feb 2011 02:14:20 PM UTC, comment #3:

Xnee 2.0 is old (5 years?) so we cannot spend time on that version.

If you update to a newer version of Xnee, is the program still there?

Can you provide us with a list of the "steps" that is recorded. Something like toe following will be fine:

1. start cnee in record mode from command line
using the following args ....

2. start program X

3. In program X, menu item C is clicked

4. Move pointer and press pointer ...

...

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Wed 09 Feb 2011 08:32:18 AM UTC, comment #2:

Thanks for the interest.

The background details you have asked for is-

OS Version- Linux version 2.6.20-1.21
X Version- X Protocol Version 11, release 6.8.2
Xnee version- Xnee 2.0

Will be happy to provide any more information you may want.

Kiran D <kiran_kdmukh>
Tue 08 Feb 2011 05:53:08 PM UTC, comment #1:

Thanks for your report,

> What would the possible cause be? How can i overcome this?


To be honest: I have no idea. But it surely it is weird enough to not attack asap. We're on it

Some more background is needed before we dive into this:

OS / DIsctribution?
X version?
Xnee version?

Henrik Sandklef <hesa>
Project AdministratorIn charge of this item.
Tue 08 Feb 2011 12:45:43 PM UTC, original submission:

I have installed CNEE and successfully able to record and replay.

I have been using cnee on an application to edit images (drag using mouse) to produce a curve. Then take a screen print of this edition.

When i replay the same and produce a replayed screen print, both the screen prints look alike.

But when pixel by pixel comparison is done for both the screen prints, it shows a difference along the border of the curve.

Surprisingly, multiple replays give a consistent curve (when replayed screen prints are compared among themselves). But when compared to the recorded screen print, all have the same difference.

What would the possible cause be? How can i overcome this?

Kiran D <kiran_kdmukh>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach File(s):
   
   
Comment:
   

Attached Files
file #22780:  snap-2.scr0.png added by kiran_kdmukh (58KiB - image/png)
file #22781:  snap-2.scr0.png added by kiran_kdmukh (58KiB - image/png)

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -unavailable- added by hesa (Posted a comment)
  • -unavailable- added by kiran_kdmukh (Submitted the item)
  • -unavailable- added by kiran_kdmukh
  • -unavailable- added by kiran_kdmukh
  • -unavailable- added by kiran_kdmukh
  •  

    Do you think this task is very important?
    If so, you can click here to add your encouragement to it.
    This task has 0 encouragements so far.

    Only logged-in users can vote.

     

    Please enter the title of George Orwell's famous dystopian book (it's a date):

     

     

    Follow 11 latest changes.

    Date Changed By Updated Field Previous Value => Replaced By
    Sun 10 Jul 2011 04:53:21 PM UTChesaStatusIn Progress=>Fixed
      Open/ClosedOpen=>Closed
    Fri 25 Feb 2011 11:54:02 AM UTCkiran_kdmukhAttached File-=>Added snap-2.scr0.png, #22780
      Attached File-=>Added snap-2.scr0.png, #22781
      Attached File-=>Added Difference-snap-2.scr0.png, #22782
    Tue 08 Feb 2011 05:53:08 PM UTChesaSeverity3 - Normal=>4 - Important
      StatusNone=>In Progress
      Assigned toNone=>hesa
    Tue 08 Feb 2011 12:45:43 PM UTCkiran_kdmukhCarbon-Copy-=>Added -unavailable-
      Carbon-Copy-=>Added -unavailable-
      Carbon-Copy-=>Added kiran_kdmukh

    Back to the top


    Powered by Savane 3.1-cleanup1