Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | merge default into bug20099bug20099 | Matthias Radestock | 2009-01-16 | 18 | -233/+603 |
|\ | | | | | | | | | | | which involved a few conflict resolutions. The most complex of these was dealing with the termination of the limiter; we now unlink from it first so we don't get a spurious EXIT signal. | ||||
| * | Merged 20071 into default | Ben Hood | 2009-01-16 | 18 | -226/+599 |
| |\ | |||||
| | * | Merged 18557 into default | Ben Hood | 2009-01-16 | 4 | -89/+390 |
| | |\ | |||||
| | | * | add some commentsbug18557 | Matthias Radestock | 2009-01-15 | 2 | -2/+6 |
| | | | | |||||
| | | * | Merged default into 18557 | Ben Hood | 2009-01-15 | 21 | -163/+1021 |
| | | |\ | |||||
| | | * | | tidying up, refactoring and some cosmetic changes | Matthias Radestock | 2008-12-26 | 4 | -61/+38 |
| | | | | | |||||
| | | * | | cosmetic: rename 'in_use' to 'volume' | Matthias Radestock | 2008-12-25 | 1 | -10/+10 |
| | | | | | |||||
| | | * | | optimisation: only notify queues that have had can_send requests rejected | Matthias Radestock | 2008-12-25 | 1 | -18/+22 |
| | | | | | |||||
| | | * | | destroy limiter when a channel becomes unlimited | Matthias Radestock | 2008-12-24 | 3 | -22/+38 |
| | | | | | | | | | | | | | | | | | | | | which results in far more efficient handling of subsequent deliveries | ||||
| | | * | | merge branch heads | Matthias Radestock | 2008-12-24 | 0 | -0/+0 |
| | | |\ \ | |||||
| | | | * | | Remove unnecessary method name from protocol_error | Tony Garnock-Jones | 2008-08-06 | 1 | -2/+1 |
| | | | | | | |||||
| | | | * | | Special-case global=true: we don't implement it | Tony Garnock-Jones | 2008-08-06 | 1 | -0/+6 |
| | | | | | | |||||
| | | * | | | don't leave the limiter behind when the channel terminates | Matthias Radestock | 2008-12-23 | 2 | -2/+13 |
| | | | | | | |||||
| | | * | | | add type spec for start_link | Matthias Radestock | 2008-12-23 | 1 | -0/+1 |
| | | | | | | |||||
| | | * | | | handle the "no limiter" case more obviously | Matthias Radestock | 2008-12-23 | 1 | -10/+8 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is no semantic change since gen_server:cast(undefined, ...) returns 'ok'. However, it only does so because it catches the 'badarg' error thrown by erlang:send. It is probably more efficient to not attempt the send in the first place. Plus for documentation purposes, and to keep dialyzer happy, it is useful to state explicitly which functions are expected to be called on an 'undefined' limiter. | ||||
| | | * | | | cosmetic | Matthias Radestock | 2008-12-23 | 1 | -20/+20 |
| | | | | | | |||||
| | | * | | | deal with limiting after consumer subscription | Matthias Radestock | 2008-12-23 | 3 | -19/+43 |
| | | | | | | |||||
| | | * | | | create limiter lazily | Matthias Radestock | 2008-12-23 | 2 | -5/+22 |
| | | | | | | | | | | | | | | | | | | | | | | | | This makes an 'unlimited' channel as efficient as it used to be | ||||
| | | * | | | make limiter keep track of all queues with subscriptions | Matthias Radestock | 2008-12-23 | 2 | -18/+52 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is more efficient since it avoids the repeated (de)monitoring and updates to the limiter state. | ||||
| | | * | | | merge default into bug18557 | Matthias Radestock | 2008-12-23 | 5 | -72/+286 |
| | | |\ \ \ | |||||
| | | | * | | | cosmetic | Matthias Radestock | 2008-12-23 | 2 | -12/+12 |
| | | | | | | | |||||
| | | | * | | | merge default into bug18557 | Matthias Radestock | 2008-12-23 | 4 | -17/+37 |
| | | | |\ \ \ | |||||
| | | | * | | | | ensure fairness | Matthias Radestock | 2008-12-23 | 1 | -4/+13 |
| | | | | | | | | |||||
| | | | * | | | | saner state transition handling | Matthias Radestock | 2008-12-19 | 1 | -52/+39 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | and assorted bug fixes | ||||
| | | | * | | | | fix bug: make sure consumers blocked due to limit are dropped from State | Matthias Radestock | 2008-12-19 | 1 | -5/+5 |
| | | | | | | | | |||||
| | | | * | | | | handle transactional acks | Matthias Radestock | 2008-12-19 | 1 | -18/+16 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | also: - simplify and optimise non-basic-get ack counting - don't talk to the limiter when there is nothing to tell, i.e. the non-basic-get ack count is zero | ||||
| | | | * | | | | minor simplifications | Matthias Radestock | 2008-12-18 | 1 | -6/+4 |
| | | | | | | | | |||||
| | | | * | | | | oops | Matthias Radestock | 2008-12-18 | 1 | -1/+1 |
| | | | | | | | | |||||
| | | | * | | | | refactoring | Matthias Radestock | 2008-12-18 | 1 | -18/+14 |
| | | | | | | | | |||||
| | | | * | | | | tidy some more comments | Matthias Radestock | 2008-12-18 | 1 | -3/+3 |
| | | | | | | | | |||||
| | | | * | | | | tidy comments | Matthias Radestock | 2008-12-18 | 1 | -11/+10 |
| | | | | | | | | |||||
| | | | * | | | | lots of tweaks and fixes | Matthias Radestock | 2008-12-18 | 1 | -48/+20 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - remove superfluous (or wrong) comments - notification and demonitoring always go together - don't change the in_use count when limit is altered - fix the limit_reached condition | ||||
| | | | * | | | | fix typo | Matthias Radestock | 2008-12-18 | 1 | -2/+2 |
| | | | | | | | | |||||
| | | | * | | | | Added handler for monitor notifications | Ben Hood | 2008-12-18 | 1 | -17/+6 |
| | | | | | | | | |||||
| | | | * | | | | Don't double monitor anything | Ben Hood | 2008-12-18 | 1 | -2/+5 |
| | | | | | | | | |||||
| | | | * | | | | Put some monitors in | Ben Hood | 2008-12-18 | 2 | -20/+26 |
| | | | | | | | | |||||
| | | | * | | | | Got rid of superfluous PD read | Ben Hood | 2008-12-18 | 1 | -2/+1 |
| | | | | | | | | |||||
| | | | * | | | | Fix for multi ack bug | Ben Hood | 2008-12-18 | 1 | -1/+1 |
| | | | | | | | | |||||
| | | | * | | | | add limiter to module list | Matthias Radestock | 2008-12-18 | 1 | -0/+1 |
| | | | | | | | | |||||
| | | | * | | | | merge default into bug18557 | Matthias Radestock | 2008-12-18 | 71 | -890/+1680 |
| | | | |\ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | and add license notice on rabbit_limiter | ||||
| | | | * | | | | | Shutting dialyzer up | Ben Hood | 2008-12-07 | 2 | -6/+17 |
| | | | | | | | | | |||||
| | | | * | | | | | Added catch for pre-fetch size | Ben Hood | 2008-12-07 | 1 | -0/+5 |
| | | | | | | | | | |||||
| | | | * | | | | | Added match for setting the global flag | Ben Hood | 2008-12-07 | 1 | -0/+4 |
| | | | | | | | | | |||||
| | | | * | | | | | Dead queue informs limiter | Ben Hood | 2008-12-07 | 3 | -1/+16 |
| | | | | | | | | | |||||
| | | | * | | | | | Merged default into 18557 | Ben Hood | 2008-12-07 | 4 | -43/+236 |
| | | | |\ \ \ \ \ | |||||
| | | | | * | | | | | Differentiate between acks for basic.get and basic.consume | Ben Hood | 2008-11-24 | 1 | -1/+13 |
| | | | | | | | | | | |||||
| | | | | * | | | | | Now the channel sends the ack directly to the limiter instead of via the queue | Ben Hood | 2008-11-24 | 3 | -10/+12 |
| | | | | | | | | | | |||||
| | | | | * | | | | | Minor fixes | Ben Hood | 2008-11-21 | 1 | -11/+15 |
| | | | | | | | | | | |||||
| | | | | * | | | | | Changed prefetch from call to cast | Ben Hood | 2008-11-21 | 1 | -6/+6 |
| | | | | | | | | | | |||||
| | | | | * | | | | | Got rid o the per-queue in-use capacity | Ben Hood | 2008-11-21 | 2 | -31/+22 |
| | | | | | | | | | |