On Wed, Mar 02, 2011 at sometime, Jay A. Kreibich wrote: > On Tue, Mar 01, 2011 at 03:21:58PM +0000, David Gilbert scratched on the wall:
> > for(jj=0; jj<tree.nDim*2; jj++){ > > sqlite3_snprintf(512-nCell,&zCell[nCell]," > > %f",(double)cell.aCoord[jj].f); > > > > That's against 3.7.4 but it looks like it's the same in the trunk. > > > > With that change the test suite passes. However, that %f looks a bit > > odd a few lines below and is probably worth thinking about. > Due to the way the floating point registers work on a PDP-7, in C all > 32-bit float values are promoted to 64-bit double values when used > in a var-arg. Hence, for output, "%f" can be safely used for both > 32-bit and 64-bit floating point values. The tags "%f" and "%lf" are > the same. (This isn't true for input functions such as scanf().) Ah yes; and indeed that's what the sqlite's own printf does - always fetches a double. Still, the %lld case in my patch does need fixing. (Note I'm not actually subscribed to the list, so please cc) Dave _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users