Of course, what's shipping in IE 8 is "broken" in that it doesn't support run to completion (and neither will Chrome 4). So honestly I'm not super compelled by the "IE shipped" argument.
I still think giving a close approximation to run to completion (repeatable reads semantics) + a callback for serializable behavior is the best corse forward at the moment. J On Tue, Dec 1, 2009 at 6:25 PM, Ian Hickson <[email protected]> wrote: > On Tue, 1 Dec 2009, Adam Barth wrote: > > > > I haven't been following the localStorage mutex discussion in detail, > > but have we already rejected the idea of having content specifically ask > > for the mutex via a transaction callback, similar to how web databases > > work? > > One of the limitations is we can't change the API (since it already > shipped in IE). If we could change the API, it'd be trivial to fix. > > -- > Ian Hickson U+1047E )\._.,--....,'``. fL > http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. > Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.' > >
