You are looking at a grid of cows for xiii

I have just had an epiphany about multiple cow databases on multiple servers.

I will write this down here so I dont forget it.

server1 does a http get to server2 which causes server2 to save a pointer record for server1 in the messages table.

time passes

server1 does another http get to server2, again causing a pointr record to be added to messages.

this go on indefinatly adding pointer records.. now the fun part is what is returned in the http message back to server1: all of the messages (global messages like shout) that have taken palce between the last 2 pointer records.

These messages are added to server1's messages table BUT any which match (by a unique ID pid thing) are not added, as they have already been seen.

Simple really.

Im thinking the http get could contain all messages server2 has not seen (so server2 manages pointers in both message tables.. this meanse server2 does not have to turn around to query server1.. it also makes the concept of daisychaining much nicer as there is only a oneway querry process.

Oh the fun of it all.

In theory, server1 will query server2 which has already queried server3 and 4 and 5 so these new messages get passed back down the linke to server1.

Ok enuf of this.. and back to work/sleep.