summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* The previous 'simplification' went a little too far, and failed to spot than ↵bug22821Matthew Sackman2010-06-071-6/+7
| | | | in queue redeclaration we don't want to permit exclusive redeclaration of a non exclusive queue
* cosmeticMatthew Sackman2010-06-071-4/+1
|
* simplify exclusive access checksMatthias Radestock2010-06-071-52/+23
|
* Don't do another dirty_read to get a queue we already have.Simon MacMullen2010-06-071-2/+14
|
* Merged in defaultSimon MacMullen2010-06-0725-103/+302
|\
| * bug22824 merged into defaultMarek Majkowski2010-06-049-27/+27
| |\
| | * 2009 => 2010bug22824Matthew Sackman2010-06-029-27/+27
| | |
| * | merge bug21932 into defaultMatthias Radestock2010-06-023-30/+224
| |\ \ | | |/ | |/|
| | * refactoringbug21932Matthias Radestock2010-06-022-36/+17
| | |
| | * Delay the issuing of channel.open_ok if under memory pressure. Also revert ↵Matthew Sackman2010-06-014-28/+48
| | | | | | | | | | | | changes to tcp_acceptor. Also add test
| | * Merging in from defaultMatthew Sackman2010-06-013-24/+217
| | |\ | |/ /
| | * cosmeticMatthias Radestock2010-05-261-2/+2
| | |
| | * improve error handling codeMatthias Radestock2010-05-261-40/+32
| | |
| | * Sorted out exception handling correctly, and added many more testsMatthew Sackman2010-05-262-49/+110
| | |
| | * cosmeticMatthias Radestock2010-05-261-2/+2
| | |
| | * improve flow control logicMatthias Radestock2010-05-261-68/+66
| | | | | | | | | | | | | | | - hopefully easier to understand - handle unexpected channel.flow-ok with an explicit error
| | * Minor refactoringsMatthew Sackman2010-05-252-13/+14
| | |
| | * Writing some tests revealed a number of mistakesMatthew Sackman2010-05-242-6/+68
| | |
| | * 1) stop accepting new connectionsMatthew Sackman2010-05-241-2/+19
| | |
| | * 3) wait for some reasonable time to receive a flow-ok; 4) if we don't get a ↵Matthew Sackman2010-05-241-54/+92
| | | | | | | | | | | | flow-ok then close the channel with an appropriate error
| | * active is much easier to think about than blocked. Deal with the possibility ↵Matthew Sackman2010-05-241-10/+25
| | | | | | | | | | | | of receiving various conserve_memory messages whilst waiting for a flow_ok to come back from a client
| | * 5) if we get a flow-ok but then still get a publish, throw away the publish ↵Matthew Sackman2010-05-241-14/+17
| | | | | | | | | | | | and close the channel with an appropriate error
| * | s/fun(/fun (/gSimon MacMullen2010-05-2814-46/+51
| | |
* | | Cosmetics.Simon MacMullen2010-05-281-2/+2
| | |
* | | Cherry-pick the queue-exclusivity-on-basic.get thing from amqp_0_9_1.Simon MacMullen2010-05-281-1/+3
| | |
* | | Cherry-pick the queue-exclusivity-on-(un)binding thing from amqp_0_9_1.Simon MacMullen2010-05-283-13/+29
|/ /
* | remove misleading commentsMatthias Radestock2010-05-281-5/+0
| |
* | cosmeticMatthias Radestock2010-05-271-4/+4
| |
* | some minor refactoringMatthias Radestock2010-05-272-82/+65
| |
* | Merge bug22596 into defaultEmile Joubert2010-05-2755-1211/+4654
|\ \
| * \ Merging bug 21824 onto defaultMatthew Sackman2010-05-275-22/+159
| |\ \
| | * | Cosmetics and formattingbug21824Matthew Sackman2010-05-274-19/+19
| | | |
| | * | Merging default into bug 21824Matthew Sackman2010-05-275-23/+160
| | |\ \
| | | * | No reason for this to be a dict of lists.Simon MacMullen2010-05-201-2/+2
| | | | |
| | | * | Forgot to add this.Simon MacMullen2010-05-201-0/+110
| | | | |
| | | * | Move the former bug21824 into a new branch that comes off bug21915 (and thus ↵Simon MacMullen2010-05-204-20/+47
| | | | | | | | | | | | | | | | | | | | default) rather than amqp_0_9_1.
| * | | | Merging bug 22812 into defaultMatthew Sackman2010-05-272-26/+36
| |\ \ \ \ | | |/ / / | |/| | |
| | * | | Minor refactorbug22812Matthew Sackman2010-05-271-3/+4
| | | | |
| | * | | We were only applying the local shortcut in the case when we were *only*Simon MacMullen2010-05-271-20/+24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | talking to a local node. This is wrong because: 1) We could then happen to pick a local delegate and a remote delegate that were the same process, and deadlock. 2) There's still a possibility of messages overtaking if sometimes they go via delegates locally and sometimes not. So fix that to always avoid the delegates when communicating locally, even if we're communicating remotely at the same time.
| | * | | Make error messages in the logs a bit clearer.Simon MacMullen2010-05-261-6/+11
| | | | |
| * | | | Merging bug21915 onto default. Note that this has some semantic changes ↵Matthew Sackman2010-05-271-2/+2
| |\ \ \ \ | | | | | | | | | | | | | | | | | | which are 'safe' for 0-8 and more in line with 0-9-1. This is: an exclusive queue can only be redeclared in the same connection as created the queue and the redeclaration must also specify exclusivity.
| | * | | | change the silent-close delay unit to secondsMatthias Radestock2010-05-271-2/+2
| | |/ / / | | | | | | | | | | | | | | | | | | | | for consistency with all the other time interval values Also rename the constant to something a little less specific
| * | | | Cosmeticsbug21915Matthew Sackman2010-05-272-33/+25
| | | | |
| * | | | Don't eat errors that could be returned by F.Simon MacMullen2010-05-261-5/+5
| | | | |
| * | | | Clean up a bit.Simon MacMullen2010-05-262-9/+3
| | | | |
| * | | | Move F into with_or_die.Simon MacMullen2010-05-261-8/+15
| | | | |
| * | | | Correct specs.Simon MacMullen2010-05-261-2/+1
| | | | |
| * | | | Merge default into 21915.Simon MacMullen2010-05-265-44/+69
| |\ \ \ \ | | |/ / /
| | * | | Merging bug 22774 onto defaultMatthew Sackman2010-05-251-3/+9
| | |\ \ \
| | | * | | Handle quoted node names by parsing with erl_scan:stringbug22774David Wragg2010-05-241-3/+9
| | | | |/ | | | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If the node name contains characters which cannot be used in a simple Erlang atom (e.g. a hyphen), then pid_to_string will quote it; e.g. <'rabbit@centos5-64'.51.0>. If in string_to_pid we try to convert the node name part back to an atom with list_to_atom, the quotes will be included in the resulting atom (e.g. '\'rabbit@centos5-64\'') and so the two functions will not be inverses. So use erl_scan to parse the node name instead, to recover an atom that exactly matches the original.
View Lorry Controller Status