On Sun, Oct 15, 2006 at 11:26:52AM +0200, Christian Ritter wrote: > 6. (already mentioned in a bug report): The (implicit) handling of some > matrix calculations is inconsistent with what Excel does. On first > sight, the Excel way is not all that "logical", but it's extremely > practical. For example, Xmat*Weight where Xmat is a nxp matrix and > Weight a nx1 vector results in WXmat where all columns of X are (scalar) > multiplied by Weight. Gnumeric does this only for the first column. The > same works when Weight is a 1xp vector. In this case the resulting WXmat > has the rows of Xmat (scalar) multiplied by weight. A third example is > the behavior of =if(vector=transpose(vector),1,0) . If the elements of > vector are different, this generates an identity matrix under Excel. > Under gnumeric, this doesn't work.
This will be patched in 1.7.2 due out tonight. _______________________________________________ gnumeric-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/gnumeric-list
