Instead of using the WTF::String(const UChar*) constructor and traversing the string to determine the length, we should use sqlite3_column_bytes16() .
Created attachment 105437 [details] Proposed patch
How much improvement?
(In reply to comment #2) > How much improvement? Over 9000.
(In reply to comment #3) > (In reply to comment #2) > > How much improvement? > > Over 9000. To clarify, I just spoke with Benjamin about this, he's concerned that the additional mutex lock (happens implicitly with any sqlite3 call) would negate any gains from avoiding the string traversal.
Comment on attachment 105437 [details] Proposed patch Clearing flags on attachment: 105437 Committed r93945: <http://trac.webkit.org/changeset/93945>
All reviewed patches have been landed. Closing bug.