Mike Kupfer wrote: > [added Mark to the cc list] > >>>>>> "Jim" == Jim Walker <James.Walker at Sun.COM> writes: > > Jim> I was talking with Mark Nelson about the clone, and he mentioned > Jim> that in the reasons they have an active clone: > [...] > Jim> 1. reduces locking issues (teamware issues) > Jim> 2. more stable than gate (ie. changed only once a day) > Jim> 3. reduces traffic on gate (ie. pulls are done most of the time) > > Jim> 2 and 3 still seem useful with Hg. > > I'm confused. Is point 3 related to system-level contention issues, or > is it something else? The parenthetical remark makes it seem like point > 3 is about how readers can starve writers with Teamware. But in that > case I don't understand the difference between points 1 and 3.
Yep. If I/O contention isn't an issue then 3 doesn't apply to Hg. Will we need to "lock" the Hg gate for big putbacks like we do now with the Teamware gate? Cheers, Jim
