It's not a deadlock though. Error codes are cheap. Why not just create another?
Also, for the record, I'm strongly against us allowing it to be possible for IndexedDB to result in a deadlock. J On Wed, Sep 22, 2010 at 8:57 PM, ben turner <[email protected]> wrote: > Hi folks, > > While implementing the latest setVersion changes I came across this > problem: > > Let's say that a site is open in two different windows and each > decides to do a setVersion request at the same time. Only one of them > can win, obviously, and the other must end up calling close() on > itself or the setVersion transaction will never run (and all database > activity will basically hang at that point). > > Jonas and I decided that this situation should result in the losing > database's request having its error handler called with DEADLOCK_ERR. > Does that sound reasonable to everyone? > > -Ben > >
