On Tue, Aug 09, 2011 at 02:54:26PM +0200, Michael Hanselmann wrote: > Am 9. August 2011 14:51 schrieb Iustin Pop <[email protected]>: > > On Tue, Aug 09, 2011 at 01:13:56PM +0200, Michael Hanselmann wrote: > >> +J_FIELDS = [ > >> + "id", "ops", "status", "summary", > >> + "opstatus", "opresult", "oplog", > >> + "received_ts", "start_ts", "end_ts", > >> + ] > >> + > > > > Mmm, I think this is too much. I understand the need for a bulk > > job/status query, but entire opstatus/opresult/oplog (note the last can > > be big)? > > > > I would restrict the bulk output here down to id, status, summary, and > > the timestamps. Anything more detailed can be requested on a per-job > > basis. > > One one hand I agree with you—it can indeed be a lot of data. On the > other hand all other resources provide the same fields for bulk > queries as for individual queries. Should we keep this consistency or > not?
Ah, good point. Hmm… one gets the feeling that we would need to add fields over RAPI :) I really don't think that getting the log, for example, at each poll is useful, so I would be evil this time and break consistency (well documented breakage, of course). What do you think? iustin
