| 00:44.09 | *** join/#storm jkakar (~jkakar@77.Red-79-150-34.dynamicIP.rima-tde.net) |
| 00:44.52 | *** join/#storm jamesh (~james@canonical/launchpad/jamesh) |
| 02:02.49 | *** join/#storm jkakar (~jkakar@77.Red-79-150-34.dynamicIP.rima-tde.net) |
| 02:51.17 | *** join/#storm jkakar (~jkakar@184.Red-88-0-173.dynamicIP.rima-tde.net) |
| 03:00.58 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 04:03.12 | *** join/#storm artista_frustrad (~artista_f@201-15-206-240.ctame704.dsl.brasiltelecom.net.br) |
| 04:24.30 | *** join/#storm jkakar (~jkakar@201.Red-79-155-137.dynamicIP.rima-tde.net) |
| 05:10.31 | *** join/#storm artista_frustrad (~artista_f@201-15-206-240.ctame704.dsl.brasiltelecom.net.br) |
| 09:59.18 | *** join/#storm danilo_ (~danilo@static-213-198-236-193.adsl.eunet.rs) |
| 11:51.48 | *** join/#storm stub (~stub@canonical/launchpad/stub) |
| 13:29.32 | *** join/#storm artista_frustrad (~artista_f@201-15-206-240.ctame704.dsl.brasiltelecom.net.br) |
| 17:14.03 | *** join/#storm skrite (~skrite@69.55.26.25) |
| 17:39.18 | midgetspy | is there a way to .find() an object only if it's in the cache? I have an object Foo that's linked to many Bar objects, and it has some custom properties that I'm manually caching that describe some things about the related Bar objects. I want to clear Foo's custom cache on Bar.__storm_invalidated__(), but if the related Foo isn't cached in Storm there's no point creating it from the DB just so I can clear its custom cache... so how can |
| 17:42.20 | jkakar | midgetspy: I don't believe there is a way to do that. |
| 17:43.00 | jkakar | midgetspy: Actually, wait. |
| 17:43.33 | jkakar | midgetspy: I've never used this feature, but ResultSet.cached looks like what you want. First do the find, then instead of iterating or otherwise accessing values yielded by the result, call .cached(). |
| 17:44.11 | jkakar | midgetspy: That said, it sounds like you probably want to implement Foo.__storm_invalidated so that Bar doesn't need to try to clean it up. |
| 17:47.33 | skrite | Does storm recommend using MyISAM or InnoDB ? |
| 17:49.37 | jkakar | skrite: You must you use InnoDB (because the other backends won't work properly with Storm). |
| 17:49.51 | skrite | jkakar, ok, cool. |
| 17:51.15 | skrite | I am getting a deadlock sometimes when i do a storm.commit() this is in a pretty busy thread. I would like to know what transaction is being done when this happens, but using the debug true for the whole thread would dump way to many sql statements to the terminal. Is there something i can do to find out what is messing things up? |
| 17:51.52 | jkakar | skrite: Are you using a Store from a different thread than which is was created? |
| 17:52.54 | skrite | not where i am seeing the failure, but it does happen in my code in places, i create a store, then pass it to an object to use when it performs some action. |
| 17:53.13 | skrite | typically, there will not be a store sent to or from another thread though |
| 17:53.35 | jkakar | skrite: The places where it happens are bugs. :) |
| 17:54.01 | jkakar | skrite: You can only safely use a store in the same thread from which it was created (this is because of the underlying drivers, not because of Storm itself). |
| 17:54.19 | skrite | ok, i will check the thread |
| 17:54.23 | skrite | thanks |
| 17:54.35 | jkakar | skrite: As far as deadlock, I'm not sure. |
| 17:54.38 | jkakar | np |
| 18:02.19 | midgetspy | is there a reason that storing something as Date() stores it as date.str() as text rather than date.toordinal() as numeric? the latter would allow SQL comparisons and sorting which are really essential imo |
| 18:02.21 | midgetspy | (in sqlite) |
| 18:02.59 | midgetspy | jkakar: well Foo.__storm_invalidated__ doesn't get called if one of the related Bars change, and that's what will affect the result of the cache (not Foo's status but the related Bars' statuses) |
| 18:04.16 | jkakar | midgetspy: To answer your first question, you can compare dates using Python: store.find(MyClass, MyClass.date < datetime.utcnow()) |
| 18:04.41 | jkakar | midgetspy: Also, I don't know about SQLite, but MySQL and PostgreSQL support string-based dates directly in SQL. |
| 18:05.31 | skrite | is there an option in commit() that allows me to see the statements that are committed ? |
| 18:06.14 | jkakar | midgetspy: __storm_invalidate__ is typically used to invalidate object-caches between transactions. I'm guess that you're manually invalidating a Bar when it changes and want to update Foo's...? Is that right? |
| 18:06.24 | jkakar | skrite: from storm.tracer import debug; debug(True) |
| 18:07.03 | skrite | :wq |
| 18:07.13 | skrite | oops, wrong terminal |
| 18:07.19 | skrite | jkakar, thanks |
| 18:07.23 | jkakar | skrite: np |
| 18:07.36 | jkakar | I'll step away for a while, good luck guys! :) |
| 18:08.38 | skrite | jkakar, thanks |
| 18:09.38 | midgetspy | jkakar: Um... say I have Library and Book, and Library has an attribute called .english_books which executes an SQL query that provides a list of english books, and caches itself. When any book that belongs to that library is invalidated I want to invalidate Library's .english_book cache. I don't want to invalidate it when the Library changes. |
| 22:16.03 | jkakar | midgetspy: Uhm. |
| 22:16.55 | jkakar | midgetspy: In that case I've guess that Book.__storm_invalidate__ would call Book.library.uncache_book(self). |
| 23:31.41 | *** join/#storm jamesh (~james@canonical/launchpad/jamesh) |
| 23:38.25 | *** join/#storm jkakar (~jkakar@193.153.247.139) |
| 23:55.08 | *** join/#storm artista_frustrad (~artista_f@201-15-206-240.ctame704.dsl.brasiltelecom.net.br) |