PR 225861 After Best fit alignment then can't get correct profile

huangyhg

After Best fit alignment then can't get correct profile

<< Junjie Zhang  --  07/19/04  09:42:08>>
I met some qeustions need to help:

Reading file to get point's data and using them to measure all points, then construction all points to a feature set(for example :Curve1). after that I used the feature set (Curve1)to build alignment with Bestfit. Final to get Curve1's profile. if I get the profile step by step by pressing menu or
button, the result is correct. But if I rerun the program ,I found the
Curve1's profile is not correct. and if I put them into loop procedure, I
found that if rerun the program more than twice , PC-Dmis will be gone.

Hope someone give me some ideas! I can provide the pcdmis program for your reference.

<<END>>


<< Changes made by Junjie Zhang -- 12/13/04  11:11:57>>
Status:  RESOLVED to CLOSED
<<END>>
<< Bret Naylor  --  11/29/04  14:49:07>>
Junjie, the report has been resolved.  Please verify that we have fixed this and if you agree, please go ahead and set the status to CLOSED.  Thanks.
<<END>>
<< Changes made by Bret Naylor -- 11/29/04  14:49:18>>
Action:  Bret Naylor to Junjie Zhang
<<END>>
<< Changes made by Junjie Zhang -- 11/29/04  16:05:19>>
Action:  Junjie Zhang to Bret Naylor
<<END>>
<< Scott Worden  --  09/09/04  15:44:32>>
wasaddedtoreadme
<<END>>
<< Bret Naylor  --  09/07/04  14:01:51>>
Reviewed.
<<END>>
<< Changes made by Bret Naylor -- 09/07/04  14:02:00>>
Action:  Bret Naylor to Junjie Zhang, Status:  REVIEW to RESOLVED
<<END>>
<< David Petrizze  --  09/02/04  13:23:58>>
Reviewed.
I see there is a bit more to the do_math method of V40 but it appears Don has covered all bases.
<<END>>
<< Changes made by David Petrizze -- 09/02/04  13:23:58>>
Action:  David Petrizze to Bret Naylor
<<END>>
<< Wade Burton  --  09/02/04  12:17:14>>
Reviewed.  I've tested Don's change with the part program here and a couple others I have and it appears to work fine for me.
<<END>>
<< Changes made by Wade Burton -- 09/02/04  12:17:24>>
Action:  Wade Burton to David Petrizze
<<END>>
<< Changes made by Don Turcotte -- 08/31/04  16:49:41>>
Action:  Don Turcotte to Wade Burton
<<END>>
<< Don Turcotte  --  08/31/04  16:49:08>>
Wade,
Could you Review the changes that were made in DIM_PROF.CPP for this PR.
<<END>>
<< Andy Roberts  --  08/31/04  13:28:28>>
Don, I didn't see a specific case where the math was not done.  It just appeared that it may not be done when editing in the edit window.  Your test confirm that it does happen.  Thanks for the verification.
<<END>>
<< Changes made by Andy Roberts -- 08/31/04  13:28:39>>
Action:  Andy Roberts to Don Turcotte
<<END>>
<< Changes made by Don Turcotte -- 08/18/04  16:52:21>>
Action:  Don Turcotte to Andy Roberts
<<END>>
<< Don Turcotte  --  08/18/04  16:51:37>>
Andy,
I tried editing the BF2D from the dialog (using F9) and also in the edit window.  I put a breakpoint on the do_math() of BF2D to verify that the math was being redone.  When editing from the dialog, the dialog OK processing will invoke executeprivate of the start align that contains the BF2D, so do_math() is called on the BF2D.  When editing from the editwin, left_the_global_group() of start_align will call executeprivate() so again the do_math() of BF2D is called.
Did you see a specific case where do_math() was not being called?
<<END>>
<< Andy Roberts  --  08/18/04  12:19:37>>
Reviewed.
Hi Don,
It looks like the math may no longer be done for a best fit 2d alignment.  If you edit the input features (add one, delete one, or change one) for a best fit 2d alignment does the math get done?
<<END>>
<< Changes made by Andy Roberts -- 08/18/04  12:20:11>>
Action:  Andy Roberts to Don Turcotte
<<END>>
<< Don Turcotte  --  08/10/04  17:27:49>>
The change to DIM_PROF.CPP should also be reviewed by Wade.
<<END>>
<< Changes made by Don Turcotte -- 08/10/04  17:00:45>>
Action:  Don Turcotte to Andy Roberts, Status:  OPEN to REVIEW
<<END>>
<< Don Turcotte  --  08/10/04  16:59:28>>
Problem with crash was call to do_math() in CPCDbf2d_align::read_text(...) which cannot be called here because the BFList may not completely set up yet.  This had already been commented out in the latest V3.5 which explains why I wasn't seeing the crash there.  The reason why V3.5 and V3.6 give different profile results was in CPCDdim_profile::do_math(...) which was checking both m_form_and_loc and m_bCreateWithoutALign and erroneously doing FORMANDLOCATION instead of FORMONLY which the user had selected.  The internal alignment option (BESTFIT or NO FIT) is new to V3.6.
Files inserted to server
------------------------
V40\ALIGNMNT\BF2D_ALI.CPP
V40\ALIGNMNT\BF3D_ALI.CPP
V40\DIMENS\DIM_PROF.CPP
V36\ALIGNMNT\BF2D_ALI.CPP
V36\ALIGNMNT\BF3D_ALI.CPP
V36\DIMENS\DIM_PROF.CPP
<<END>>
<< Don Turcotte  --  08/09/04  17:49:11>>
Rewriting the program to avoid using a set of a set, eliminates the crash, but the profile results for V3.5 and V3.6 are different.  The crash in BF2D involes the way BFItem's handle search id's.  Still working on figuring this out...
<<END>>
<< Don Turcotte  --  08/03/04  14:03:52>>
This works in the latest build of V3.5.  After execution Prof4 = 0.1072, Prof5 = 0.1521.  Repeated executions produce the same result without crashing.  In latest V3.6, I fixed code in BF3D_ALI.CPP (not uploaded yet) so won't crash with null pointer.  First execution with V3.6 gives Prof4 = -0.5211, Prof5 = 0.5419.  Second execution will crash ap.  This is related to search_id's for constructed sets, but I haven't figured out what the critical difference between V3.5 and V3.6 is yet.
<<END>>
<< Don Turcotte  --  08/02/04  17:45:58>>
I am able to duplicate this problem.  The crash is caused by CPCDbf3d_align::BFListFromCmd(...) call to pCmd = ((CPCDsearch_id*)pCmd)->get_searched_command(); returning null.  I don't yet fully understand what is going wrong with the CBFItem lists.
<<END>>
<< Changes made by Tim Wernicke -- 07/27/04  09:18:39>>
Action:  Tim Wernicke to Don Turcotte, Status:  MOREINFO to OPEN, Priority:  to Medium
<<END>>
<< Changes made by Junjie Zhang -- 07/27/04  08:37:29>>
Action:  Junjie Zhang to Tim Wernicke
<<END>>
<< Tim Wernicke  --  07/26/04  15:45:38>>
Junjie, no files are attached.
<<END>>
<< Changes made by Tim Wernicke -- 07/26/04  15:45:40>>
Action:  Tim Wernicke to Junjie Zhang
<<END>>
<< Junjie Zhang  --  07/26/04  08:21:49>>
This is attachment for you reference. the file name is profileTest.rar
<<END>>
<< Changes made by Junjie Zhang -- 07/26/04  08:22:34>>
Action:  Junjie Zhang to Tim Wernicke
<<END>>
<< Tim Wernicke  --  07/18/04  21:09:47>>
Please attach the example part program, point data file and probe files(s).
<<END>>
<< Changes made by Tim Wernicke -- 07/18/04  21:09:59>>
Action:  Don Turcotte to Junjie Zhang, Status:  OPEN to MOREINFO
<<END>>