Thank you M. Medcalf for your nice explanation.
In my first post, I gave half of the used solution : when storing a
"truncated value", eventual remaining digits are allso, separately,
stored as a whole integer.
Both parts are reassembled later when needed (i.e. when doing set
agregation). The intent is to (imperfectly) recover lost precision
resulting from truncation at elementary level : naîve and ugly, but the
best I'm able to do.
From your explanation, can I conclude that when doing summation on
large sets of fp numbers, a prior sort on this set gives chances to a
better accuracy of the result ? In SQL, if the usage of a separate
table to be filled in sorted order is not wanted, what would be the
differences in efficiency/memory usage between a sub-select in the FROM
clause compared to the same ordered set processed via a WITH clause ?
Thanks again.
-jm
---
L'absence de virus dans ce courrier électronique a été vérifiée par le logiciel
antivirus Avast.
https://www.avast.com/antivirus
_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
