View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0001845||CCdciel||General||public||17-11-14 00:40||17-11-19 15:12|
|Priority||normal||Severity||major||Reproducibility||have not tried|
|Summary||0001845: Trunk 643, exception errors|
|Description||Got this night a few times exception errors which stopped operation until acknowledged. Will investigate tomorrow.|
|Tags||No tags attached.|
||Did a quick check but could not find anything. Will run next nightly session in the debugger to trace the exceptions.|
I get several runerror(201) in line 635 in fu_starprofile. See attached. The variables are all fine see watches.
For the rest it is all running fine! Slippage correction works fine. Only the blue background of the sequence I can't hardly see behind my red front. Making bold?
Error 201 is "range check error" , this appear only in debug and it is trapped by the try...except I add to every function that access the image array. But this must be fixed.
In this row the only thing that can raise a range check is: img[0,j,i0+i]
One possibility is the image frame is smaller than 150.
There is no test in this function that FStarX +/- s/2 is in the image frame because it is alway called after gethfd that do this test.
Yes the active row in the sequence use fixed color and this is not good, I try with bold.
||rev 658 use bold font instead of color to mark the current row.|
I could read in the debugger the value of img[0,j,i0+i] so that is strange. However in simulation I can also read img[0,j,i0+i+100000].
Maybe it has something to do with an occasional drift, I observe during focusing. The focus star sometimes drift near the edge of the window . The scope just moved to the guide star and gears are still settling. This also matches with the unpredictable occurrence.
Adding this to the code img[0,j,i0+i+100000] gives the same error. Running this without the debugger gives an "Range Check Error"
A try & exception or min & max would prevent this error.
||If drift is the problem during focusing, the other thing you could consider what to do if the star is too close to the border. Beyond there will be an abort since there is a low snr value but if it is partly beyond this could give an wrong focus result. It is very rare so should probably not considered. It would require a check if the measuring box hits the borders.|
When the program is compiled in Release mode, reading img[0,j,i0+i+100000] will not raise an exception if the address is still in the process memory. If this is out of the allowed range the try...except show a message in the log but the program is not aborted.
When compiled in Debug mode the compiler add the range checking code to test for this kind of error and the exception is always raised.
If the star is too close to the border the FindStarPos() function safely fail and the auto-focus fail. Same if the star go out of the window.
The solution in this case is to increase the frame size in Preference/Focus.
In revision 660 I add the testing for a valid range and eventually adjust the histogram width when the star is too near the border but still detected by FindStarPos().
|17-11-14 00:40||han||New Issue|
|17-11-14 11:55||han||Note Added: 0004274|
|17-11-18 21:28||han||File Added: error ccdciel0.png|
|17-11-18 21:28||han||File Added: error ccdciel1.png|
|17-11-18 21:28||han||Note Added: 0004289|
|17-11-18 22:20||Patrick Chevalley||Note Added: 0004290|
|17-11-18 22:31||Patrick Chevalley||Note Added: 0004291|
|17-11-19 11:17||han||Note Added: 0004292|
|17-11-19 11:25||han||Note Edited: 0004292||View Revisions|
|17-11-19 11:28||han||Note Edited: 0004292||View Revisions|
|17-11-19 15:02||han||Note Added: 0004293|
|17-11-19 15:11||Patrick Chevalley||Note Added: 0004294|
|17-11-19 15:12||Patrick Chevalley||Status||new => resolved|
|17-11-19 15:12||Patrick Chevalley||Resolution||open => fixed|
|17-11-19 15:12||Patrick Chevalley||Target Version||=> 1.0|