This is a public chat log generated from the #semsol IRC channel.
07:53:07
hi sam
12:16:24
kwijibo: when using LIMIT 1000 OFFSET 0, how to you ensure the order of the triples does not change?
12:17:05
scor: that's a good point, i forgot to mention that
12:17:11
I suppose you can't expect the triples to be returned in the same order, or does ARC guarantees it
12:17:13
?
12:17:26
maybe adding a SORT would be good
12:17:28
i think you just have to repeat until you don't get any triples back
12:18:36
scor: you can add a SORT, but I think most triple stores won't return a random order
12:18:55
but users might be adding /removing triples while you're migrating the data
12:18:57
I know, but what if in the second query, the order is different, and the triple which appeared at the position 1001 appears at the position 234 and therefore not in the window LIMIT 1000 OFFSET 1000
12:19:38
that's a different issue
12:20:03
even with a fixed dataset, I wonder if a stable sorting can be guaranteed
12:21:28
scor, for the second part, we didn't run the next query until we'd made sure we changed the first batch
12:21:39
actually, in the second part, we didn't page
12:21:47
we just run with a limit
12:21:52
until we've finished
12:22:22
(ie, until there are no bad triples left)
12:22:38
my description in the blog post is misleading
12:22:45
i'll change it
14:35:34
how old was kier when he started learning to code? 8?
14:35:38
bloody auto channel switching crappy IRC client
14:37:18
kwijibo wonders if kier's ported jena to assembler yet ...