FYI, Robert tells me the p2p-hackers list is (justifiably) badmouthing P2's concurrency control... I guess we're this close to being able to say we have a...
Folks, I will be unfortunately unable to attend the P2 meeting on Monday due to the looming NSDI deadline on Tuesday. Also I will be out of town for SOSP on...
The type casts not longer work. I looked at oper.[hC] and discovered much of it is commented out. Do we have another way of supporting type conversion? Thanks,...
Operators automatically cast the result to the type of the left operand. Prior typecasts were not exposed to the language, so at some point we took this...
I've been having some difficulty in writing rules that end up being delta rules. Consider: programCompletion programComplete(@A, Pid) :- rulesComplete(@A, Pid,...
Folks, I've set up an audio bridge and a web sharing thingy for the P2 Monday group meeting: Monday, October 22, 2007, 09:30 AM US Pacific Time 916-356-2663,...
Hi Tyson, Why not use the (perhaps naive) approach to have delete events resulting in a deletion in the head? I don't quite see what problem might occur apart...
I think it boils down to dealing with duplicate base relation tuples. If you delete a base relation tuple whose values are not unique across a join defined by...
mmm, my previous post is naive indeed. However, if it is only the problem of plural matching tuples, how about simply remove them all and issue a warning...
This is just a thought but Tuple lineage might solve the problem. If I delete a particular tuple in table 1 then joining that deleted tuple with table 2 would...
Correct, the old planner does a DELETE_ACTION for a DELETE_EVENT in the delta rule. INSERT_ACTION is used for REFRESH_EVENT, since reinserting a tuple...
So there's a whole slew of papers talking about multiply derived tuples in such views. The "delete and recomputed" approach seems to be the state of the art....
This is why for hard-state tables, the old planner was using the reference counted implementation: to deal with multiply derived tuples. For soft-state tables,...
My memory must be going. The reference count does indeed do the trick. All I need to do is make sure specific delta event types (delete, insert, refresh) do...
I can't seem to rid myself of type issues regarding the compareTo routine. If we must continue with the type lattice then perhaps we should remove all of these...
I'd love to eradicate int32, uint32 and uint64. Mothy has taken a first step in his branch ("scheduler", which Robert is currently debugging) but the type...
I just fixed integer types in pel... going through pel converts all integer values to 64 bits. As you can imagine this really messes things up when doing index...
I'm afraid I can't give a good estimate yet of how long it will take me to debug. The code in the scheduler branch says that it is passing all 55 unit tests,...
So that I can make it to Sigmod... We should talk about a more permanent solution after. ... So that I can make it to Sigmod... We should talk about a...
OK, agreed, thanks! ________________________________ From: p2group@yahoogroups.com [mailto:p2group@yahoogroups.com] On Behalf Of Robert Soule Sent: Thursday,...
My point was, would pushing everything up to 64bit integers not fix the problem in the same way? If not, then this is a reasonable workaround until after...
We'll have the meeting as usual. Monday, October 29, 2007, 09:30 AM US Pacific Time 916-356-2663, 8-356-2663, Bridge: 4, Passcode: 1863314 Live Meeting:...
Folks, Here is the bridge info for today's meeting: Monday, November 05, 2007, 09:30 AM US Pacific Time 916-356-2663, 8-356-2663, Bridge: 2, Passcode: 9690249 ...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
