On 27 Oct 2011 at 22:04, Kees Nuyt <k.n...@zonnet.nl> wrote: > On Thu, 27 Oct 2011 21:47:17 +0100, Simon Slavin > <slav...@bigfraud.org> wrote: > >> >> On 27 Oct 2011, at 8:12pm, Tim Streater wrote: >> >>> Is there a way to get a unique id from a database >>> without actually creating a new row? >> >> Sure. Do something like >> >> SELECT max(id) FROM mytab; >> >> Then in your own code, add 1 to it and use that as >> the "id" for the row you're about to save. >> Make sure you handle the NULL case (where mytab >> doesn't have any rows in it yet) correctly. > > And wrap it all in an IMMEDIATE or EXCLUSIVE transaction, or > you'll get a race condition if some other process tries to do the > same at the same time. > All processes have to behave, if there's one that does the SELECT > above in uncommitted mode and proceeds to use it, you have a > problem. > So, better use AUTOINCREMENT.
Thanks for the various responses. I hadn't thought of the max(id) approach but I'm not sure I can put a BEGIN/COMMIT around everything, as there are network exchanges involved there too. Nothing at the SQLite web site gave me a better clue than my own approach, either, but it doesn't look like there's anything obvious I overlooked. -- Cheers -- Tim
_______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
