OK, makes sense to me. > -----Original Message----- > From: Babich, Alan [mailto:ABabich@filenet.com] > Sent: Thursday, June 24, 1999 5:49 PM > To: 'Jim Whitehead' > Subject: RE: JW6: return 102 if long time > > > In my experience, systems don't even try to guess > at how long a query will take to execute. Estimating > a single user on a fast network is not feasible, > let alone multiple users, varying server load, > and varying network capacity and traffic. > > Of course it would be nice to predict end-to-end query > execution time, but I don't think it is > feasible. To time out after N seconds > and tell the client something he already knows > (N seconds have gone by) is unnecessary complexity > to REQUIRE in my opinion. > > Alan Babich > > -----Original Message----- > From: Jim Whitehead [mailto:ejw@ics.uci.edu] > Sent: Thursday, June 24, 1999 5:02 PM > To: 'DASL' > Subject: RE: JW6: return 102 if long time > > > > > > I don't want to require that. I don't see that the benefit > > is enough to require that things be more complicated. > > I still think this would be nice to have, but it may be that > search engines > typically don't know how long a search will take ahead of time. > Is this the > case? Or are most searches typically quite fast, a la AltaVista? I'm not > aware of, e.g., typical DM performance for searches. > > - Jim >Received on Thursday, 24 June 1999 20:55:16 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 22 March 2009 03:38:04 GMT