On Dec 16, 8:53 am, Filipe wrote:
> What's the clever way to do this?
Well, probably the easiest way would be to compute the difference
between the UNIX timestamps (in seconds) and do a setTimeout with a
timeout value of that difference (multiplied by 1000 to convert to
milliseconds).
--
Job Bo
If you care about failure resilience at all:
1. You will need a pseudoqueue in SQL (or whatever) with a datetime field
indicating when the event needs to fire.
2. Every, say, 5 minutes (settimeout) query SQL for anything that needs to
happen in the next, say, 6 minutes.
3. Calculate the differen
Could be done in redis also, with a sorted set, where the values are json
payloads of task data and the scores are timestamps. On an interval (like
Jonathan suggested) you could do a query for all items older than 'now' and
start processing them.
On Sunday, December 16, 2012 12:12:11 PM UTC-8,
On 12/16/12 12:47 PM, mscdex wrote:
On Dec 16, 8:53 am, Filipe wrote:
What's the clever way to do this?
Well, probably the easiest way would be to compute the difference
between the UNIX timestamps (in seconds) and do a setTimeout with a
timeout value of that difference (multiplied by 1000 to