----- Original Message ----- From: Flemming Videbaek To: brat-dev-l@bnl.gov Sent: Saturday, April 14, 2001 6:37 PM Subject: Brat update Indeed several errors were found in the BRAT tracking code ,and it is in fact a bit amazing at least these has not shown up before. Most errors are related to unitialized variables so maybe this may be related to have not run into this before. a) ROOT 3.00.06 requires the methods Compare to be Compare(const Tobject..) const these were added as CH has done earlier for some other classes. There are still a few left which I did not fix (try cd brat; grep Compare */inc/*.h to see if the proper methods are there. b) The BrTPCClusterFinder did not set the limits for data to be cleared in ClearWorkingArea. This would mainly affect early events since after a while presumably the whole 2d array would be cleared.This was essentialy two typo's c) BrLocalTrack() did not set fCovar=0, and would leave this unitialized and at some point a delete was attempted. The code crashed. In addition fNhits can be 2, and there was a statement chisq/(fNHits-2) which is not so good. In BrLocalTrack the error calculation added in Feb is bypassed. The errors are already calculated in the covariance matrix. the local members fPosErrror[i]... has been set to sqrt(covar[][]) approbiately. d) After these changed a version of BramSelectMRS ran apperently ok on pii6. e) The brat version was bumped to 1.15.3 Happy Easter ------------------------------------------------------ Flemming Videbaek Physics Department Brookhaven National Laboratory tlf: 631-344-4106 fax 631-344-1334 e-mail: videbaek@bnl.gov
This archive was generated by hypermail 2b29 : Sun Apr 15 2001 - 13:10:27 EDT