On 11/10/15, Paul Sanderson <sandersonforensics at gmail.com> wrote: > > As the last thing I did was a wal-checkpoint then according to the > documentation the page should have been written to the DB (it was see > pic below of page two of the DB) AND the salts incremented and > randomised (section 4.3 of the file format page). >
The increment and randomization of the salts occurs at the start of the next write transaction. (It is deferred until then to avoid having to take unnecessary write locks.) The file format documentation is not technically wrong, but it is misleading. I have updated that document now (see the second paragraph of section 4.3 at https://www.sqlite.org/fileformat.html) to try to make it less ambiguous. -- D. Richard Hipp drh at sqlite.org
