Hi all,

 I implemented both the bug fixes found in the last days: right charge
correlation for modes with B+-> D0->Kspipi and rejection of events with
wrong B-D flavor correlation for charged B's.

 I checked the impact of this on data and MC and on the final fit.

 First let's see the effect of rejecting events with the wrong B-D flavor
correlation. In the following plots you can find the comparison of the mes
distribution before (black) and after (red) this rejection:

 Bch GENERIC MC

 http://www.slac.stanford.edu/~daniele/vub/gene.os

 Bch DATA

 http://www.slac.stanford.edu/~daniele/vub/data.ps

 (the B0's are obviously unchanged and not shown)

 As you see the effect and the gain in the purity are large. For generic
MC the background removal is larger (since it is just bbar background).

 Once I applied these changes I refitted the data.
 Here the model is generic+crossfeed

 DEFAULT

 BRBR = 0.0170771 +- 0.00430631(stat) +- 0.00239891(MC stat)

 NEW VALUE (after changes)

 BRBR = 0.0141994 +- 0.00421423(stat) +- 0.00234686(MC stat)

 All the difference comes from the background modeling, as you can see
from the following plots that show the comparison of the Mx distribution
before (black) and after (red) the bug fixes.

 data

 http://www.slac.stanford.edu/~daniele/vub/test.ps

 Vcb model

 http://www.slac.stanford.edu/~daniele/vub/vcbcomp.ps

 Data are basically unchanged. Differences are in the vcb model. The tail
atlow Mx is larger. The result is that now the fit on Bch goes down

 BRBR = 0.0237497 +- 0.00520536(stat) +- 0.00270644(MC stat)

 before was 0.028

 Daniele