<     May 2017     >
Su Mo Tu We Th Fr Sa  
    1  2  3  4  5  6  
 7  8  9 10 11 12 13  
14 15 16 17 18 19 20  
21 22 _2_3 24 25 26 27  
28 29 30 31
00:18 bturker joined
00:40 codenamedmitri joined
01:19 bturker joined
01:41 irclogger_com joined
01:41 Topic for
02:57 bturker joined
03:58 bturker joined
04:58 bturker joined
05:59 bturker joined
06:10 bturker joined
06:24 churcho_ joined
06:26 DaJoe joined
07:10 bturker joined
07:56 bturker joined
08:56 bturker joined
09:24 bturker joined
09:28 Siyo joined
09:28 Siyo joined
10:25 bturker joined
10:39 bturker joined
11:28 bturker_ joined
12:29 tombriden joined
12:39 hive-mind joined
13:58 churcho joined
13:59 churcho joined
14:00 andyt1 joined
14:00 churcho joined
14:01 churcho joined
14:02 churcho joined
14:03 churcho joined
16:46 daxorid left
17:16 <MitchellSalad> can you delete a data type with the KV api?
18:01 kleptocroc- joined
18:57 enqae joined
20:01 <MitchellSalad> i have determined you can, although if a modification happens concurrently, you'll end up with a sibling...
20:01 <MitchellSalad> so my strategy is basically to keep re-deleting in a tight loop, updating the vector clock each time, until I see that the key is gone
20:01 <MitchellSalad> (to avoid sibling explosion)
20:03 <MitchellSalad> is this ill-advised? also, is there any documentation on deleting a data type via the KV api? I just played around with a riak instance manually to learn how it behaves
21:13 bturker joined
21:22 <russelldb> MitchellSalad: yup, sorry, all we have the is kv api for deletes, deletes are tricky and we didn't really tackle them
21:22 <russelldb> and that sibling thing is a known issue
21:23 <russelldb> https://github.com/basho/riak_kv/issues/1360
21:24 <russelldb> I think if riak was a great big map, then a delete concurrent with an update…the update should win?
21:28 <MitchellSalad> ah, okay. I'm a little confused why the siblings appear, but otherwise the CRDT seems to behave normally. the tombstones just pile up, if you never happen to peek at the data type using the KV api?
21:28 <MitchellSalad> is it always the case that only one sibling is the actual data type, and the rest are tombstones?
21:43 <russelldb> MitchellSalad: should be the case, yes
21:43 <russelldb> the merge code checks each sibling, and if it is a data type, it will merge it
21:44 <russelldb> I'd say it needs some work
21:46 <MitchellSalad> cool, thank you sir
22:13 MitchellSalad_ joined
22:17 MJD_ joined
22:20 pigmej_ joined
22:26 <MitchellSalad_> is riak_kv still under active development?
22:42 leifw joined
22:42 supermike_ joined
22:42 yo61 joined
22:42 alexmoore joined
22:42 sam1_ joined
22:44 yo61 joined
23:02 MJD joined