On 2017/12/20 11:01 AM, Dominique Devienne wrote:
"extremely efficient" is a bit exaggerated IMHO.
More space efficient definitely, see serial types 8 and 8 in [1], so only
the record's header section need to be consulted,
w/o the need to decode the whole record or decode the varint [2], thus more
cpu efficient too (since less to do), but the
record's page still needs to be paged in/out on reads/writes, and that's
like to dominate anyway. Just my guess though,
Not to mention that unless you remake the table (which the OP was
insistent to avoid), that INT column will be at the very end of a list
of much bigger columns, which would make reading it the single most
inefficient column to read in the whole table, although the subsequent
index that you are surely to add on that Column will undoubtedly be one
of the fastest, second only to the rowid.
no hard data to back that up. Perhaps Ryan will measure the difference he
often does :) --DD
LOL, I would, but a quick glance at the importantometer for this result
revealed that the exercise will be scheduled right between "Some other
time" and "Never". :)
Cheers!
Ryan
_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
