PR# 259469 V44 Dimension/GDT Tests: There seems to be significant delays in

huangyhg

V44 Dimension/GDT Tests:   There seems to be significant delays in calculating profile of these scans.
<< Don Ruggieri  --  02/03/09  09:08:48>>
data is in 02032009_02a.zip.  There seems to be significant delays in calculating profile of these scans.
Steps to reproduce:
I am working on testing simultaneous evaluations of profile dimensions.  I go to the end of this program and added Linear profile dimension of two scans, one at a time.  The delays in processing each of these dimension commands are nearly 20 seconds.  Then, when I do the simultanoues evaluation, it exceeds 30 seconds.
I am not sure if this is warranted because of of the calculations required, or if it is an unnecessaryy delay of some kind.
<<END>>

<< Changes made by Neil Kay -- 02/24/09  18:57:25>>
Status:  CLOSED to RESOLVED
<<END>>
<< Changes made by Neil Kay -- 02/24/09  18:57:12>>
Status:  RESOLVED to CLOSED
<<END>>
<< David Petrizze  --  02/18/09  14:20:09>>
Reviewed.
<<END>>
<< Changes made by David Petrizze -- 02/18/09  14:20:24>>
Action:  David Petrizze to Don Ruggieri, Status:  REVIEW to RESOLVED
<<END>>
<< Don Turcotte  --  02/09/09  08:15:53>>
I have uploaded my fixes as described below.
Files inserted to server
------------------------
V44B\DIMENS\DATDEF.CPP
V44B\DIMENS\DimSimultaneousDlg.CPP
V44B\DIMENS\FEATCTRLFRM.CPP
V44B\DIMENS\SIMULTANEOUS_EVALUATION.CPP
V44B\FCF\GEOMETRICALTOLERANCE.CPP
V44B\INCLUDE\DimSimultaneousDlg.H
V44B\INCLUDE\SIMULTANEOUS_EVALUATION.H
V44B\INCLUDE\XACTGEOMETRYMATH\minimization.H
V44B\INCLUDE\XACTGEOMETRYMATH\OrientationFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\OrientationPlanarFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\PositionCylFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\PositionFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\PositionPlanarFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\PositionSphereFunc.H
V44B\INCLUDE\XACTGEOMETRYMATH\SurfProfileFunc.H
V44B\XACTGEOMETRYMATH\minimization.cpp
V44B\XACTGEOMETRYMATH\SurfProfileFunc.CPP
V44B\NETDLL\XactGeometryMath.dll
V44B\NETDLL\XactGeometryMathD.dll
V44B\NETLIB\XactGeometryMath.lib
V44B\NETLIB\XactGeometryMathD.lib
<<END>>
<< Changes made by Don Turcotte -- 02/09/09  08:16:25>>
Action:  Don Turcotte to David Petrizze, Status:  OPEN to REVIEW
<<END>>
<< Don Turcotte  --  02/06/09  17:10:47>>
I have completed my testing and fixes for this.  Everything seems to be working correctly now.  I cannot upload yet because the update server is down for the switch to Mercurial.  I plan to upload my changes Monday morning.
<<END>>
<< Don Turcotte  --  02/05/09  17:14:56>>
I have completed testing of my changes for F9 editing/creating of simultaneous evaluation command.
I have found that the do_math(...) of the simultaneous evaluation command does the fit of all the data repeatedly for each FCF in the list.  This is unnecessary and time consuming.  I have fixed this so the fit of all the data is done only once and the results applied to each FCF in the list.
I have also made a change so that the math will not be done on program load.  This seems to work ok for the attached program.  I am concerned that we should be doing the math on program load for OPTIMIZED_FIT since the call to do_math(...) from the serialize(...) method of the simultaneous evaluation command was added for the OPTIMIZED_FIT development.  Therefore, I have made an additional change to call do_math(...) on program load only for OPTIMIZED_FIT.  This should not be a time problem in this case since the optimized fit processing uses the last saved solution as a seed for the math.  So the do_math(...) on program load for OPTIMIZED_FIT should be relatively fast.  Still needs more testing.
<<END>>
<< Don Turcotte  --  02/04/09  17:28:24>>
I have not made any progress on eliminating the one call to do_math in the simultaneous evaluation commands on program load.
However, I was able to duplicate the problem of F9 on the simultaneous evaluation command, then Close followed by a long delay.  I also found other such problems related to F9 editing on the simultaneous evaluation command.  These problems are all related to unnecessary calls to do_math when nothing in fact has been changed.  I have fixed these problems but I am still testing my changes.
<<END>>
<< Don Turcotte  --  02/03/09  17:07:18>>
Loading the "a" program in Release build takes about 3:31 minutes on my computer.  This is a problem of do_math(...) being called on load.
If the profile FCF's are fully constrained with Datums A,B,C (using LIN1 for B, and PNT6 for C), then the evaluation is very fast because there is no fitting so slowness is a result of fitting algorithm.
I did not see problem of F9 on simultaneous evaluation command then Close.  For me this closed the dialog immediately.
Found problem with creating new DATDEF command.  The causes math of following FCFs to be redone even though they don't use the new datum.  Therefore, closing DATDEF dialog is very slow while we wait for math to complete.  I have fixed this.
<<END>>
<< Changes made by Tim Wernicke -- 02/03/09  09:30:39>>
Priority:  to Critical
<<END>>
<< Don Ruggieri  --  02/03/09  10:03:58>>
Having removed the profile dimensions and the simultaneous evaluations at the end of the program, it now opens much faster.
<<END>>
<< Don Ruggieri  --  02/03/09  09:52:32>>
More data is in 02032009_02b.zip.  In this case, just opening the program leads to a delay of nearly two minutes.  This program is the same as the original, with perhaps a couple of additional commands.  I am suspicious that the delays are being caused by the same issue.
<<END>>
<< Don Ruggieri  --  02/03/09  09:30:09>>
Another example of a similar delay is this.  I re-open this program and go to  the end. Then I place the cursor on one of the Simultaneous Evaluation commands and press F9 to open the dialog.  Without any further action, I press the Close button.
It takes about 20 seconds to simply close the dialog.  I am not trying to modify it nor evaluate anything, just close the dialog.  Why the delay?
<<END>>

huangyhg

else// if (pFeature->type() != POINTCLOUD_OPER) // Anis - 257270, PR259553