| 01:26.22 | *** join/#storm milosn (~milosn@cable-178-149-0-183.dynamic.sbb.rs) |
| 03:42.18 | *** join/#storm nicedice_ (~nicedice@unaffiliated/nicedice) |
| 06:21.39 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 06:46.36 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 07:29.51 | *** join/#storm evilaliv3 (~evilaliv3@nat-cataldo.sssup.it) |
| 08:26.22 | *** join/#storm milosn (~milosn@cable-178-149-0-183.dynamic.sbb.rs) |
| 10:25.58 | *** join/#storm fcorrea_ (~fcorrea@nat/canonical/x-drpwyzvecsapihtn) |
| 13:52.56 | *** join/#storm dajac (~dajac@254-113.63-188.cust.bluewin.ch) |
| 15:08.59 | magmatt | emhs: I've not done that. I'm not sure there are hooks readily available. What are you trying to do? |
| 16:10.52 | *** join/#storm _mup_ (~mup@li37-90.members.linode.com) |
| 16:13.03 | *** join/#storm nicedice (~nicedice@unaffiliated/nicedice) |
| 17:11.40 | *** join/#storm jkakar (~jkakar@nat-204-14-239-210-sfo.net.salesforce.com) |
| 18:08.50 | *** join/#storm evilaliv3 (~evilaliv3@93-46-232-189.ip109.fastwebnet.it) |
| 19:11.53 | *** join/#storm nicedice (~nicedice@unaffiliated/nicedice) |
| 19:59.06 | evilaliv3 | hey jkakar =) |
| 19:59.28 | evilaliv3 | i'm looking for the information regarding an old fix and probably an existing memory leak |
| 19:59.33 | evilaliv3 | Bug fixes |
| 19:59.34 | evilaliv3 | --------- |
| 19:59.34 | evilaliv3 | <PROTECTED> |
| 19:59.34 | evilaliv3 | <PROTECTED> |
| 19:59.46 | evilaliv3 | this is a fix annovered in the 0.16 release |
| 20:00.11 | evilaliv3 | here is the bug i've probably spotted: |
| 20:00.14 | evilaliv3 | https://github.com/globaleaks/GlobaLeaks/issues/611#issuecomment-24108841 |
| 20:00.33 | evilaliv3 | i don't understand why returning a list from a transact makes a leak |
| 20:01.00 | evilaliv3 | i've tryed also by doing an copy.deepcopy(list) but with no success |
| 20:10.49 | evilaliv3 | ops nothing jkakar i've found my bug, not storm one |
| 21:36.29 | jkakar | evilaliv3: Ah, nice! |
| 21:37.54 | evilaliv3 | =) |
| 21:59.14 | evilaliv3 | jkakar: i've a question for you related to storm |
| 21:59.34 | evilaliv3 | suppose i've a transaction that need to make atomic operation involving db and filesystem |
| 22:00.08 | evilaliv3 | having filesystem operation inside the transaction is bad as they can be long (suppose storing a 1GB file) |
| 22:00.26 | jkakar | evilaliv3: Go on. |
| 22:00.48 | evilaliv3 | how can this accomplished with storm ? |
| 22:00.56 | evilaliv3 | (i'm using sqlite) |
| 22:01.09 | evilaliv3 | currently i'm using two different transaction like this: |
| 22:01.23 | evilaliv3 | 1) read needed data from db with transact_1 |
| 22:01.29 | evilaliv3 | 2) make filesystem operation |
| 22:01.49 | evilaliv3 | 3) if all has gone good make a transact_2 committing data |
| 22:01.58 | evilaliv3 | and so registering file operation in db |
| 22:02.09 | evilaliv3 | but doing this way i'm strongly not atomic |
| 22:02.34 | evilaliv3 | what do you suggest in this case? |
| 22:04.51 | evilaliv3 | the problem is known in literature, but i miss how to accomplish this with storm |
| 22:04.55 | evilaliv3 | (http://stackoverflow.com/questions/2422603/how-to-manage-transaction-for-database-and-file-system-in-java-ee-environment) |
| 23:02.58 | *** join/#storm milosn (~milosn@cable-178-149-0-183.dynamic.sbb.rs) |
| 23:04.14 | *** join/#storm fcorrea (~fcorrea@194.168.195.98) |
| 23:08.25 | *** join/#storm milosn (~milosn@cable-178-149-0-183.dynamic.sbb.rs) |
| 23:43.25 | jkakar | evilaliv3: Hmm, interesting. |
| 23:43.46 | jkakar | evilaliv3: Off the top of my head, I guess that what you really want is something like two phase commit, but that's a bit of a PITA. |
| 23:44.11 | jkakar | evilaliv3: Could you get away with: (1) write the file to a temporary location, (2) run the database transaction, (3) move the file to the permanent location…? |
| 23:45.05 | jkakar | Of course, #3 could fail (ie, disk full, etc.) and then you could be in trouble, but maybe you can store some information about that kind of inconsistency (or have it implicit in the way your store the files/write the transaction) to make cleanup easy/possible…? |
| 23:48.15 | jkakar | evilaliv3: I'm a bit fuzzy on the details now, it's been awhile since I looked at it all, but I think you can hook into the transaction package to do what you need. |
| 23:49.04 | jkakar | evilaliv3: If you look in storm/zope/zstorm.py you can see how we register the store with the transaction package. You can write/register a FilesystemTransactionManager and use two-phase commit to build something that will actually work. |
| 23:53.40 | *** join/#storm milosn (~milosn@cable-178-149-0-183.dynamic.sbb.rs) |