Dataserver data loss?
Can someone please clarify what this means:
"If more data is requested using the same query_id before the dataserver event dealing with that query_id has resolved the data will be trashed."
Since it isn't possible to choose the query ID used then aren't the chances of this happening incredibly remote? Also, what exactly does it mean by trashed? Is the implication that if I have a pending dataserver event with id 1234, that if a second event were to be queued with the same ID (somehow) then it would replace the existing one, preventing me from ever seeing the first event?
-- Haravikk (talk|contribs) 04:32, 12 July 2012 (PDT)
- Without looking at the article, I would assume what they meant to say was something to the effect of, if you store the request UUID in a single global key instead of in a global list of requests, if you make more than one request, you will overwrite the key and when the original request comes through you won't be able to identify it because you have overwritten the key. I don't know. -- Strife (talk|contribs) 21:17, 15 July 2012 (PDT)
- That's a possibility, but if so then I'm not sure of the value of including it, or it should at least be re-worded as an advisory on how best to handle more than one data server query as currently it reads a lot like some kind of bug in the feature itself. I think I'll do that, but I'll move the original comment here.
- "If more data is requested using the same query_id before the dataserver event dealing with that query_id has resolved the data will be trashed. To avoid the trashing do not ask if(query_key == query_id). Though (see above caveats) this has its own drawbacks."