On 6 Jun 2011, at 5:30pm, Gabor Grothendieck wrote: > I and others use SQLite from R and R itself gives Inf, -Inf and NaN > for 1/0, -1/0 and 0/0 respectively so it would reduce the differences > between the database and R if it worked in the same way. Perhaps an > option could control this behavior so that backwards compatibility > could be maintained.
While I understand the wish for better support of IEEE-754, I think it might be wise to go the whole way rather than make a single improvement. Ideally some sort of maths compatibility PRAGMA which could be set in legacy/IEEE/SQL mode. Eventually we could probably lose the legacy mode. Compatibility with SQL mode is probably just as important as compatibility with IEEE, by the way. Plenty of people seem to prototype their app with SQLite on a standalone computer before moving to a multi-user system hosted by another implementation of SQL. So having one SQL act the same as the other is important too. It's just a shame that the SQL and IEEE standards are mutually incompatible. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
