Refactor Application class #154
Conversation
| scheduleAt(simTime() + exponential(0.00001 * myAddress), pk); | ||
| break; | ||
| } | ||
| return; |
There was a problem hiding this comment.
we know that no more to do something in this method, so I wrote this to return.
we should keep the scope small to make it more readable.
| if (traffic_pattern == 2) { | ||
| int endnode_dest_addr = getOneRandomEndNodeAddress(); | ||
| EV_INFO << "My connection setup request will be sent from " << my_address << " to " << endnode_dest_addr << "\n"; | ||
| ConnectionSetupRequest *pk = createConnectionSetupRequest(endnode_dest_addr, number_of_resources); |
There was a problem hiding this comment.
I don't want to think about the detail of the packet creation.
so I move this into createConnectionSetupRequest method
…tion::handleMessage the router sends the packet to ConnectionManager, not to Application. this part never works, so I removed it.
zigen
force-pushed
the
refactoring-application
branch
from
9038538
to
b873509
Compare
|
Oh, and ultimately this is related to what the API for applications ought to be; what's the quantum equivalent of a BSD socket? That's an actual research issue that has been on the list for several years, and no one has really pursued it in depth yet. Wojtek's observation that "time is part of the service" is definitely relevant here -- the API should include high-precision timestamps, probably below the nanosecond level for some purposes. |
I refactored
Applicationclass