| 00:00.23 | *** join/#storm _mup_ (~mup@li37-90.members.linode.com) |
| 01:24.17 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 04:07.22 | *** join/#storm bigdog (~scmikes@8.8.197.72) |
| 05:55.44 | *** join/#storm lifeless (~robertc@122.58.129.196) |
| 07:21.31 | *** join/#storm skeezix (~skeezix@2002:d8eb:84e:0:de9c:52ff:fe07:b2be) |
| 08:13.06 | *** join/#storm bigjools (~quassel@canonical/launchpad/bigjools) |
| 08:32.40 | *** join/#storm skeezix (~skeezix@2002:d8eb:84e:0:de9c:52ff:fe07:b2be) |
| 12:49.03 | *** join/#storm stub (~stub@125.24.137.151.adsl.dynamic.totbb.net) |
| 12:49.03 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 13:30.38 | *** join/#storm niemeyer (~niemeyer@201-11-202-175.pltce701.dsl.brasiltelecom.net.br) |
| 13:30.38 | *** mode/#storm [+o niemeyer] by ChanServ |
| 16:39.22 | *** join/#storm keppla (~keppla@p4FD46D56.dip.t-dialin.net) |
| 16:57.55 | *** part/#storm stub (~stub@canonical/launchpad/stub) |
| 20:26.18 | magmatt | Why is the default isolation level for postgres serializable? (I'm not complaining, I want to know the reason) |
| 20:50.18 | radix | magmatt: two reasons, I think |
| 20:50.29 | radix | magmatt: one is that the users were all using serializable in their applications |
| 20:50.41 | radix | magmatt: two is that it allows for the best correct caching |
| 20:50.57 | radix | actually, I should say "easiest" correct caching |
| 20:51.05 | magmatt | radix: so I may take a performance hit moving to read-committed? |
| 20:51.33 | magmatt | oh, easiest |
| 20:52.41 | radix | the first point is really the trump, though |
| 20:53.02 | radix | we used serializable, and so we developed storm to support it very well :) |
| 20:53.22 | radix | magmatt: you won't take a performance hit, no |
| 20:53.57 | radix | magmatt: Using serializable means that you know exactly when you need to invalidate your caches; if you're using read committed, you basically can't cache correctly |
| 20:54.12 | radix | but if you're using read committed, you probably don't *want* to cache correctly |
| 20:54.54 | magmatt | right... so storm doesn't introduce any uncertainty that isn't already present with read-committed, correct? |
| 20:55.04 | magmatt | that would be a bug |
| 20:57.01 | radix | I think it's a matter of which cache you use |
| 20:57.05 | radix | it's parameterized |
| 20:57.22 | radix | (it's been a really long time since I've used storm, sorry :) |