summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* merge from defaultAlexandru Scvortov2011-01-1910-247/+341
|\
| * merge bug23705 into defaultMatthias Radestock2011-01-180-0/+0
| |\
| * | sync_timer => sync_timeoutMatthew Sackman2011-01-182-8/+8
| | |
| * | Specialise the sync_timer in order to permit us to unset the sync_timer_refMatthew Sackman2011-01-182-5/+13
| |/
| * merge bug23631 into defaultMatthias Radestock2011-01-186-58/+99
| |\
| | * Merging heads of defaultMatthew Sackman2011-01-183-9/+13
| | |\
| | | * Correct specsMatthew Sackman2011-01-183-9/+13
| | | |
| | * | Fix surprisingly long-lived mistake.Simon MacMullen2011-01-181-1/+1
| | |/
| | * Merging bug23698 into defaultMatthew Sackman2011-01-183-34/+45
| | |\
| | * | InliningMatthew Sackman2011-01-181-5/+3
| | | |
| | * | A little more abstraction never hurtMatthew Sackman2011-01-181-23/+16
| | | |
| | * | Some might say this is a touch over abstracted. It's quite pretty thoughMatthew Sackman2011-01-181-6/+27
| | | |
| | * | don't send routing-failure-induced confirms straight awayMatthias Radestock2011-01-171-5/+6
| | | | | | | | | | | | | | | | to allow for coalescing with preceding confirms.
| | * | prioritise confirms and coalesce them until we see another commandMatthias Radestock2011-01-171-12/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | As well as sending confirms when the channel goes idle we also send them on the first non-confirm (Erlang) message. I.e. coalescing is restriced to adjacent confirms in the channel's mailbox. By prioritising the 'confirm' erlang messages we shuffle all confirms to (near) the top of the mailbox, adjacent to each other. With this change we get our 'confirms will be sent eventually' guarantee back. And performance appears to remain unchanged.
| | * | coalesce confirms in channel, for better performanceMatthias Radestock2011-01-171-8/+22
| | | | | | | | | | | | | | | | | | | | We accumulate confirms in the channel and only send them when the channel goes idle.
| * | | cosmeticMatthias Radestock2011-01-181-3/+2
| | | |
| * | | correct specsMatthias Radestock2011-01-181-3/+2
| | | |
| * | | Added a test which will, timing dependent, hit the close_fds callback stuff. ↵Matthew Sackman2011-01-171-0/+30
| | | | | | | | | | | | | | | | Code coverage in tests bares this out
| * | | Merging default into bug23631Matthew Sackman2011-01-172-7/+8
| |\ \ \
| * | | | refactor: extract helper for initialising test msg store clientMatthias Radestock2011-01-161-16/+12
| | | | |
| * | | | cosmeticMatthias Radestock2011-01-161-4/+4
| | | | |
| * | | | cosmetic: get rid of intermediate var and save some spaceMatthias Radestock2011-01-161-8/+7
| | | | |
| * | | | make recover_index_and_client_refs do what it saysMatthias Radestock2011-01-161-8/+8
| | | | |
| * | | | cosmeticMatthias Radestock2011-01-161-10/+8
| | | | |
| * | | | some cosmetic and minor refactoring for clarityMatthias Radestock2011-01-161-22/+25
| | | | |
| * | | | refactor: extract helper fun for updating the pending confirm mappingMatthias Radestock2011-01-161-43/+35
| | | | | | | | | | | | | | | | | | | | and invoking the MsgOnDiskFun client callback
| * | | | simplifying refactorMatthias Radestock2011-01-161-13/+11
| | | | |
| * | | | merging default into bug23631Matthew Sackman2011-01-153-155/+59
| |\ \ \ \
| * | | | | Whoops - client_terminate should not clear the cref, unlike ↵Matthew Sackman2011-01-151-4/+3
| | | | | | | | | | | | | | | | | | | | | | | | client_delete_and_terminate. Tests pass again now.
| * | | | | Introduce msg_store_client_init/3 helperMatthew Sackman2011-01-141-7/+8
| | | | | |
| * | | | | Rearrangement in gc. Can't seem to get it shorter than this thoughMatthew Sackman2011-01-141-7/+8
| | | | | |
| * | | | | change close_all_indicated to return {ok, CState}Matthew Sackman2011-01-142-6/+6
| | | | | |
| * | | | | Correct order of function argumentsMatthew Sackman2011-01-141-5/+5
| | | | | |
| * | | | | Combine client_refs with client_msg_on_disk_callback into clientsMatthew Sackman2011-01-141-75/+65
| | | | | |
| * | | | | And now it worksMatthew Sackman2011-01-142-5/+7
| | | | | |
| * | | | | Minor tweaks, and added explanatory comments relating to the various racesMatthew Sackman2011-01-142-26/+27
| | | | | |
| * | | | | More testing needed, but this does seem to workMatthew Sackman2011-01-144-63/+136
| | | | | |
* | | | | | merge from defaultAlexandru Scvortov2011-01-185-153/+67
|\ \ \ \ \ \ | | |_|_|_|/ | |/| | | |
| * | | | | Or notSimon MacMullen2011-01-181-1/+0
| | | | | |
| * | | | | Merge bug23689 into default.Simon MacMullen2011-01-184-33/+27
| |\ \ \ \ \
| | * | | | | small simplifying refactorMatthias Radestock2011-01-181-19/+15
| | | | | | |
| | * | | | | cosmeticMatthias Radestock2011-01-181-4/+3
| | | | | | |
| | * | | | | cosmeticMatthias Radestock2011-01-181-4/+1
| | | |_|/ / | | |/| | |
| | * | | | Every msg in a queue should have a different guid. SighMatthew Sackman2011-01-171-6/+7
| | | |_|/ | | |/| |
| | * | | allow hibernation after confirmMatthias Radestock2011-01-171-1/+1
| | | |/ | | |/|
| * | | indicate absence of config files more clearlyMatthias Radestock2011-01-181-1/+1
| | | |
| * | | oopsMatthias Radestock2011-01-161-2/+0
| | | |
| * | | display config file locations on startupMatthias Radestock2011-01-162-2/+22
| |/ /
| * | Merge bug23675Simon MacMullen2011-01-141-8/+17
| |\ \
| | * \ Merge with defaultRob Harrop2011-01-141-8/+17
| | |\ \ | | | |/ | | |/|
View Lorry Controller Status