[rabbitmq-discuss] weird timeout declaring queue
jon at jbrisbin.com
Thu Jan 27 12:39:14 GMT 2011
On Jan 27, 2011, at 5:40 AM, Matthew Sackman wrote:
> On Wed, Jan 26, 2011 at 07:28:43PM -0800, John DeTreville wrote:
>> Jerry and I spoke with Jon today, and we agreed that this is a hard area to work in. I'm sending him a few very simple, special-purpose new queue implementation of mine (rabbit_ram_queue, rabbit_mnesia_queue, and the upcoming rabbit_disk_queue) to try building on. This might be easier for the moment than building on the more complicated rabbit_variable_queue, as Jon is doing now.
> Ahh, yes. It's an astonishingly bad idea to try and modify vq - that
> beast is complex. The backing_queue interface is hopefully well enough
> documented that it should be possible to build directly from there - the
> specs are there too after all.
> If the documentation is not good enough for this purpose then we should
> fix that.
It would be great to see a flowchart or something of expected input and output (that's easier to grok than the specs). The specs are helpful, but inadequate to fully understand what's going on because there's no explanation of why something is important or what the reason is of doing it that way. Like sending back a queue length. A doc explaining why that's important and how its used would be better for the developer than trying to follow it all the way through the code and infer it from how its being used.
I do think some documentation beyond what's currently available is needed if there's any expectation of community involvement on this front. New persisters to different backing stores in the future, etc...
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rabbitmq-discuss