| 00:36.41 | tjs | who is stuart bishop? |
| 00:38.21 | tjs | who claims storm has been designed to work with 'billion row tables' |
| 00:39.10 | tjs | some of those statements sound fairly fluffy |
| 00:58.46 | Fujitsu | I think he's probably best-placed to state that. |
| 00:58.52 | Fujitsu | He's the Canonical DBA. |
| 00:59.21 | Fujitsu | tjs: ^^ |
| 00:59.24 | tjs | :) |
| 00:59.43 | tjs | well, Id suggest rewording a little, my first impression was 'sales pitch' |
| 02:23.22 | radix | eh |
| 02:23.30 | radix | storm will handle any size database that your database can handle |
| 04:50.02 | *** join/#storm jbot (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 04:50.02 | *** topic/#storm is The Storm Python ORM - http://storm.canonical.com/ |
| 04:56.41 | niemeyer | tjs: It's not too hard to write it, but it's tricky |
| 04:57.07 | niemeyer | tjs: The transaction part, mostly |
| 04:57.32 | niemeyer | Because it's not very obvious how to join the running transaction |
| 04:58.12 | tjs | niemeyer: yeh, I just wrote some helper functions a week ago to allow us to spawn threads in zope3 for long running tasks, that sets up the whole zodb/site/util etc context |
| 04:58.52 | tjs | well, how do you join the running transaction? define a custom IDataManager? |
| 04:59.00 | tjs | or something |
| 04:59.37 | tjs | or is that top secret ? ;) |
| 04:59.46 | niemeyer | No.. there are a few different entry points for that |
| 04:59.50 | tjs | ah |
| 05:00.18 | niemeyer | It's not secret... but it's not a one-liner because of the way things generally work. |
| 05:00.31 | niemeyer | We don't use the same approach used by DBAPI wrappers for instance |
| 05:03.09 | tjs | ok |
| 05:06.02 | tjs | niemeyer: niemeyer: why not implement an IDataManger for storm though? it seems like the place to interact with a transaction in a pluggable way |
| 05:06.07 | tjs | oops |
| 05:14.40 | niemeyer | tjs: Actually |
| 05:14.48 | niemeyer | tjs: I just lied unintentionally.. |
| 05:15.04 | niemeyer | tjs: I intended to say what I said afterwards |
| 05:15.19 | niemeyer | tjs: It's not using the same mechanism used by normal DBAPI wrappers I've seen |
| 05:15.27 | niemeyer | tjs: It does need an IDataManager |
| 05:15.40 | niemeyer | (I think anything integrating with the transaction will need it) |
| 05:17.40 | tjs | thats what I was going to do for my integration layer, if zstom is going to take a while |
| 05:18.33 | tjs | so your IDataManager is registered for storm based content, and does commit/rollback in the appropriate methods? |
| 05:18.50 | niemeyer | Right |
| 05:18.52 | tjs | the bit I have yet to figure out is the thread-local store access |
| 05:19.08 | tjs | I guess you register a utility that returns a new store? |
| 05:19.09 | niemeyer | That's the easy part IMO |
| 05:19.16 | tjs | are stores thread-safe? |
| 05:19.34 | niemeyer | They're not at the moment |
| 05:19.36 | tjs | or you just open multiple stores, one per transaction/time the utility is called? |
| 05:19.38 | tjs | ok |
| 05:19.43 | niemeyer | But it doesn't even matter in that context |
| 05:20.08 | niemeyer | The store is bound to a connection, and thus a transaction |
| 05:20.27 | tjs | ah |
| 05:20.32 | niemeyer | tjs: Nah.. it was fine :) |
| 05:20.36 | tjs | :) |
| 05:20.41 | niemeyer | tjs: Thanks for answering |
| 05:20.48 | tjs | np |
| 05:21.08 | Fujitsu | tjs: I would have been a lot more violent if I had answered, I think |
| 05:21.15 | tjs | Fujitsu: hehe |
| 05:22.26 | tjs | Fujitsu: I'm an ubuntu user ;) woop woop woop . |
| 05:22.40 | Fujitsu | Good to know! |
| 05:22.46 | tjs | Fujitsu: whats your interest in storm btw? :) |
| 05:22.54 | tjs | just curious |
| 05:23.57 | tjs | niemeyer: so you just call create_database() for each transaction? |
| 05:24.19 | tjs | Fujitsu: doing zope stuff as well? |
| 05:25.08 | Fujitsu | Well, it's a Canonical thing that has been recently opened. That in itself is interesting, but I deal with a fair bit of Zope, and other database-driven Python (and, unfortunately, PHP) applications at work. |
| 05:25.42 | tjs | ah cool |
| 05:26.04 | Fujitsu | I, like a lot of Ubuntu devs, are interested in the supposed eventual LP open-sourcing. |
| 05:26.35 | tjs | that would be nice |
| 05:27.11 | tjs | I'm interested in seeing LP become a more wholistic solution for OSS projects |
| 05:27.56 | tjs | It's going in the right direction, but to setup a project I still need to find dedicated hosting to run things like automated testing etc.. |
| 05:28.52 | tjs | also, a wiki would be nice |
| 05:29.16 | baijum | I was going through mailing list, what is "thread-local" ? any pointer ? |
| 05:29.41 | tjs | in relation to zope? |
| 05:29.51 | tjs | or in general? |
| 05:30.06 | baijum | in general |
| 05:31.19 | tjs | http://docs.python.org/lib/module-threading.html see 'class local' |
| 05:31.37 | tjs | "Thread-local data are data whose values are thread specific." |
| 05:33.07 | baijum | tjs, thanks ! |
| 05:33.15 | tjs | baijum: np :) |
| 06:19.43 | niemeyer | Hmm |
| 06:19.57 | niemeyer | I seem to be missing some mails in the mailing list |
| 06:20.11 | niemeyer | I don't know if they're going over private conversations or if there's some issue with my mail (or the ML) |
| 06:20.33 | tjs | niemeyer: Which ones? |
| 06:20.59 | niemeyer | tjs: Stuart has answered two mails for which I don't have the originals |
| 06:21.45 | tjs | yeh I replied to someone via Stuart's email, then realised the mail I was responding to had never actually made it to the list |
| 06:22.21 | niemeyer | Ah, ok.. so it's either a private mail or a ML issue.. probably the first |
| 06:24.43 | tjs | niemeyer: re your last post |
| 06:25.09 | tjs | thats why you would only ever do a db.commit on a storm db in a data manager |
| 06:25.14 | tjs | not manually |
| 06:25.56 | tjs | that way all calls to commit will only happen once every data manager agrees they are happy with the transaction |
| 06:28.25 | niemeyer | tjs: So |
| 06:28.40 | niemeyer | tjs: You're talking about the transaction mechanism, which is fine, ok |
| 06:28.57 | niemeyer | tjs: But whatever *hooks* into that transaction mechanism, must support the two-phase strategy as well |
| 06:29.27 | niemeyer | tjs: Right? |
| 06:29.41 | tjs | by whatever you mean 'something that is a database or data layer' ? |
| 06:30.05 | *** join/#storm jace (n=root@gateway2.comat.com) |
| 06:30.24 | tjs | if so what it needs to provide is commit and rollback feature |
| 06:30.38 | tjs | which storm does :) |
| 06:30.42 | niemeyer | tjs: Ok.. let me try to illustrate a reason why this isn't the case |
| 06:30.48 | tjs | ok |
| 06:31.03 | niemeyer | tjs: sqlite3 has commit/rollback |
| 06:31.18 | niemeyer | tjs: Now.. imagine you've got two connections |
| 06:31.25 | niemeyer | tjs: To two different databases |
| 06:31.29 | niemeyer | (e.g. files on disk) |
| 06:32.05 | niemeyer | tjs: Zope's transaction asks everyone to "vote" on the commit going on |
| 06:32.16 | niemeyer | tjs: Everyone flushes, and it's fine |
| 06:32.29 | niemeyer | tjs: Cool.. it means we can commit |
| 06:32.40 | niemeyer | tjs: Now the transaction system says, ok, everyone can commit now |
| 06:32.49 | niemeyer | tjs: The first commit works |
| 06:32.55 | niemeyer | tjs: The second one fails, for whatever reason |
| 06:33.01 | niemeyer | tjs: How do you rollback the first one? |
| 06:33.26 | tjs | well.. you cant |
| 06:33.40 | tjs | but Id argue that someone voted incorrectly ;) |
| 06:33.47 | niemeyer | tjs: Precisely |
| 06:34.01 | niemeyer | tjs: That's why two-phase commit needs *support* from whatever hooks into the transaction system |
| 06:34.15 | niemeyer | tjs: Not just commit/rollback |
| 06:34.28 | tjs | ok |
| 06:35.01 | tjs | so that support is in what form? the ability to revoke an already commited transaction? |
| 06:35.05 | niemeyer | tjs: So that the voting has more meaning than just "I *think* I'm good to go", and then break on a conflict with another transaction |
| 06:35.46 | niemeyer | tjs: Databases that support it have a "prepared commit" stage that "guarantees" that the transaction can go in |
| 06:35.52 | tjs | ah |
| 06:36.46 | tjs | which dbs have that? |
| 06:37.29 | tjs | that storm supports |
| 06:38.22 | niemeyer | tjs: Postgres has it |
| 06:38.52 | niemeyer | tjs: It was added recently |
| 06:39.22 | niemeyer | tjs: http://www.postgresql.org/docs/8.2/static/sql-prepare-transaction.html |
| 06:40.36 | tjs | so storm supports that? |
| 06:41.33 | tjs | its kinda important as we will be mix-and-matching zodb/storm objects, with references |
| 06:41.48 | tjs | (although I hope not for to long) ;) |
| 06:42.55 | tjs | aah |
| 06:43.50 | tjs | niemeyer: I guess in my data manager I could do a store.execute("prepare transaction 'foo';") |
| 06:44.41 | tjs | etc |
| 06:48.00 | niemeyer | tjs: Yep, it works fine |
| 06:48.18 | tjs | well thats cool |
| 06:48.32 | Fujitsu | Are you meant to be able to break an object by changing the value of its primary key? |
| 06:49.24 | niemeyer | Fujitsu: I'm not sure I understand the question |
| 06:49.29 | niemeyer | Fujitsu: You can change the primary key |
| 06:49.51 | Fujitsu | niemeyer: I changed the primary key of one of my objects, and rolling back didn't affect it. |
| 06:49.53 | niemeyer | Fujitsu: I don't get the point about being able to break an object.. do you *want* to break the object? |
| 06:50.03 | Fujitsu | Upon committing, the object wouldn't do much. |
| 06:50.14 | Fujitsu | Threw exceptions when accessing attributes, etc. |
| 06:50.33 | niemeyer | Fujitsu: Well.. that's something else, and I'm actually curious about what you think the right behavior would be |
| 06:50.53 | Fujitsu | I'm not sure what the right behaviour would be. I just thought it a little odd (though somewhat understandable) that it did that. |
| 06:51.16 | niemeyer | Fujitsu: What should Storm do with that object? Should it continue linking the object with the new primary key? |
| 06:52.05 | niemeyer | Fujitsu: or should it consider the prior primary key to be this object.. |
| 06:52.10 | niemeyer | Fujitsu: It's mostly a caching issue.. |
| 06:52.23 | Fujitsu | Noted. I'm not sure what the correct behaviour should be. |
| 06:52.32 | niemeyer | Fujitsu: The behavior changed along Storm's life.. the problem is harder from a conceptual perspective than from an implementation one |
| 06:53.57 | niemeyer | Fujitsu: You can change primary keys, even then.. |
| 06:54.24 | niemeyer | Fujitsu: I think we'll probably figure out a behavior that matches what implementations expect with time |
| 06:55.31 | Fujitsu | niemeyer: Right, it's still early days. Is this even used in any LP code yet? |
| 06:55.49 | niemeyer | Fujitsu: You mean, changing primary keys? |
| 06:55.59 | Fujitsu | No, storm itself. |
| 06:56.21 | Fujitsu | Changing primary keys is likely a corner case that shouldn't ever be used. I just did it to see if it exploded or not. |
| 06:56.26 | niemeyer | Fujitsu: Yes.. we've ported most of it early this year |
| 06:57.00 | Fujitsu | So it is pretty well tested. Good to see. |
| 06:57.04 | niemeyer | Fujitsu: It's not in production yet, though |
| 06:57.15 | niemeyer | Fujitsu: But there is another project which uses only Storm |
| 06:57.23 | niemeyer | Fujitsu: So yeah, there's a good amount of real-world testing |
| 06:57.47 | Fujitsu | Landscape, perhaps? |
| 06:57.57 | niemeyer | Fujitsu: Besides the comprehensive test suite, of course.. |
| 06:57.57 | tjs | radix' thingo I think |
| 06:58.04 | niemeyer | Fujitsu: Yop |
| 06:58.32 | tjs | niemeyer: which project group are you part of? lp? |
| 06:58.44 | niemeyer | Yeah.. radix, jkakar, michelp, bigkevmcd, and myself work on the same project |
| 06:58.54 | tjs | ah |
| 06:59.38 | tjs | I really enjoyed working with radix :) |
| 07:00.15 | tjs | I guess he is less of a programming hobo now |
| 07:24.45 | tjs | night |
| 07:31.58 | *** join/#storm shredwheat_ (n=pete@cpe-76-168-41-58.socal.res.rr.com) |
| 07:32.15 | shredwheat_ | storm.exceptions.ClassInfoError: <type 'type'>.__storm_table__ missing |
| 07:32.42 | shredwheat_ | Trying to work my way outwards to find what this problem means... |
| 07:32.48 | Fujitsu | shredwheat_: You need to have __storm_table__ defined in your class. |
| 07:33.06 | shredwheat_ | from "store.add(User)". The database does have a 'user' table. and the class does have a __storm_table__ attr. |
| 07:33.08 | shredwheat_ | :-( |
| 07:33.25 | Fujitsu | You need to create the user table in the DB, then. |
| 07:33.38 | shredwheat_ | is there a way to get storm to list the tables it thinks are in the db? |
| 07:34.01 | Fujitsu | I'm not sure. There should be one for each class in the store. |
| 07:35.11 | shredwheat_ | I'm wondering if it is this older MySql version 4.0.x. |
| 07:35.19 | shredwheat_ | The older version was causing SQLAlchemy some trouble. |
| 07:36.10 | Fujitsu | I don't think that should be a problem. Have you actually got a `user' table in the database? |
| 07:36.51 | shredwheat_ | yes |
| 07:37.04 | shredwheat_ | that is the only table in there currently. |
| 07:37.18 | Fujitsu | Oh, I see you said that up there. Oops. |
| 07:37.31 | shredwheat_ | i'll take a look into the storm src here.. |
| 07:37.38 | Fujitsu | <type 'type'>... hm... Can you pastebin your class definition? |
| 07:38.11 | shredwheat_ | http://pastebin.com/d6ed8a0b0 |
| 07:38.18 | shredwheat_ | Python 2.4 |
| 07:40.29 | Fujitsu | Works for me on an in-memory SQLite. |
| 07:41.15 | Fujitsu | b. |
| 07:41.17 | Fujitsu | *brb |
| 07:43.08 | shredwheat_ | bigger picture.. http://pastebin.com/d5aa54076 |
| 07:43.24 | shredwheat_ | shoot, no sqlite on this server |
| 07:45.40 | Fujitsu | Back. |
| 07:46.09 | shredwheat_ | just added a traceback to the codebin... http://pastebin.com/d4d54287 |
| 07:47.02 | Fujitsu | Ah. |
| 07:47.17 | Fujitsu | You need to add an object, not a class. |
| 07:47.34 | shredwheat_ | ah. hence the <type> |
| 07:48.30 | Fujitsu | Yep, though I didn't immediately guess it meant that. |
| 07:51.35 | shredwheat_ | Slow start for me... store.find(User, User.uid=="tester").one() |
| 07:51.41 | shredwheat_ | TypeError: Expected str, found <type 'unicode'>: u'tester' |
| 07:54.46 | shredwheat_ | well switching Chars() to Unicode() and searching with u"tester" does work. |
| 07:55.45 | Fujitsu | Not sure about that. |
| 07:55.55 | niemeyer | Hey guys |
| 07:55.57 | Fujitsu | It could be MySQL's charsets doing strange things. |
| 07:56.02 | Fujitsu | Hi again, niemeyer. |
| 07:56.10 | shredwheat_ | Guess i'll go with all unicode. That's what I see in the tutorial anyways. |
| 07:56.13 | niemeyer | shredwheat_: Oops.. let me read the backlog.. one moment |
| 07:56.27 | Fujitsu | Unicode is a Good Thing, usually. |
| 07:59.31 | shredwheat_ | this is cool stuff. a bit of magic mojo going on :-) |
| 07:59.59 | niemeyer | shredwheat_: store.add() takes instances |
| 08:00.11 | niemeyer | shredwheat_: Oh, that's been said I guess, sorry |
| 08:00.52 | shredwheat_ | niemeyer, that one seems obvious in hindsight. |
| 08:01.33 | niemeyer | shredwheat_: VARCHAR in MySQL is unicode indeed |
| 08:01.49 | shredwheat_ | ah, then everything fits. |
| 08:02.01 | niemeyer | shredwheat_: Well, it's not *unicode*, to be precise |
| 08:02.13 | niemeyer | shredwheat_: It's textual data, encoded as requested |
| 08:02.25 | Fujitsu | niemeyer: What would happen if one were to create an object with a Reference to another, and store the former (but not latter) in the DB? |
| 08:02.33 | niemeyer | shredwheat_: Storm handles that as unicode instances |
| 08:03.06 | niemeyer | Fujitsu: Storm will add the whole hierarchy |
| 08:03.43 | Fujitsu | OK, so it is `viral'. Nice, nice. |
| 08:03.55 | niemeyer | Fujitsu: It is.. you can even do it after the fact |
| 08:04.00 | niemeyer | Fujitsu: I mean.. something like this works: |
| 08:04.08 | niemeyer | obj1.ref = obj2 |
| 08:04.12 | niemeyer | store.add(obj1) |
| 08:04.15 | niemeyer | <PROTECTED> |
| 08:10.22 | shredwheat_ | does the store connect the database at initialization? or is it lazy? |
| 08:14.53 | shredwheat_ | Also, does the store do a commit when destructed? or a rollback? |
| 08:25.30 | niemeyer | shredwheat_: The store connects it when initialized |
| 08:25.47 | niemeyer | shredwheat_: When destructed it'll try to close the connection only |
| 08:26.05 | shredwheat_ | so the commit is my responsibility |
| 08:30.08 | niemeyer | shredwheat_: Right, it either has do be done manually or by the framework |
| 08:30.27 | shredwheat_ | almost got this running on my site. liking it. |
| 08:30.39 | niemeyer | shredwheat_: In cases where you don't do it manually with other systems it means the framework is doing it for you |
| 08:30.59 | niemeyer | shredwheat_: Storm may be injected in the same hooks so that the automatic commit happens as well (or the rollback, for that matter) |
| 08:30.59 | shredwheat_ | yeah, i've used django a bit. but that requires the manual save on objects. |
| 08:31.35 | niemeyer | shredwheat_: So you have to "checkpoint" each object manually? |
| 08:32.03 | shredwheat_ | Looks like Store is subclassable. Looks like anyone can cook to taste. |
| 08:35.43 | niemeyer | shredwheat_: It is indeed |
| 08:43.06 | shredwheat_ | Know what MySQL data type is really stored as strings? |
| 08:45.35 | shredwheat_ | VARBINARY? |
| 08:48.15 | niemeyer | shredwheat_: Hmm.. I don't remember.. |
| 08:48.21 | shredwheat_ | got it, working now. |
| 08:48.58 | shredwheat_ | using VARCHAR but adding the BINARY attribute |
| 08:50.39 | niemeyer | shredwheat_: Ah, cool |
| 08:50.58 | shredwheat_ | thanks for your help guys. and congrats on storm. looks good. |
| 08:51.11 | niemeyer | shredwheat_: I think it'd be nice to have a table with Storm types => different databases types over time |
| 08:51.21 | niemeyer | Hmmm |
| 08:51.23 | shredwheat_ | if you guys have facebook accounts you can see it in action. heh. |
| 08:51.26 | niemeyer | Actually, it'd be good to have it soon |
| 08:51.46 | niemeyer | shredwheat_: Oh, cool |
| 08:51.51 | niemeyer | shredwheat_: Where can I see it? |
| 08:52.39 | shredwheat_ | http://www.facebook.com/apps/application.php?id=2411199646 |
| 08:52.59 | shredwheat_ | Pretty minimal use of a database at this point. But using storm is way nicer than MySQLdb. |
| 09:28.37 | shredwheat_ | thanks for help again guys. past bedtime. |
| 09:50.27 | *** join/#storm dialtone (i=dialtone@conference/europython/x-5b557b16935372e3) |
| 09:55.16 | *** join/#storm rayo (n=rayo@gw.ptr-80-238-129-209.customer.ch.netstream.com) |
| 09:56.27 | baijum | One suggestion about tutorial |
| 09:57.26 | baijum | If you have docutils installed in the web server where you running MoinMoin, you can use rst as the text format of tutorial.txt |
| 09:58.20 | baijum | then, at the beginning of that file you can add : #format rst |
| 10:02.06 | niemeyer | baijum: Oh, nice |
| 10:14.39 | niemeyer | Lunch time.. bbiab |
| 10:14.43 | baijum | niemeyer, I hope you know the rst directives for contents |
| 10:14.58 | baijum | if not: |
| 10:14.59 | baijum | .. contents:: |
| 10:14.59 | baijum | .. sectnum:: |
| 10:36.27 | *** join/#storm tjs (n=tjs@CPE-124-181-153-235.vic.bigpond.net.au) |
| 11:16.18 | *** join/#storm tjs (n=tjs@CPE-124-181-153-235.vic.bigpond.net.au) |
| 11:54.02 | *** join/#storm apoirier (n=kvirc@LAubervilliers-151-11-27-221.w193-251.abo.wanadoo.fr) |
| 12:06.30 | *** join/#storm creiht (n=cthier@nat-vlan0201.sat4.rackspace.com) |
| 13:48.21 | *** join/#storm rbriski (n=rbriski@c-67-180-47-102.hsd1.ca.comcast.net) |
| 14:08.16 | *** join/#storm niemeyer (i=niemeyer@conference/europython/x-4a83c21519074b6a) |
| 15:23.44 | *** join/#storm brosner (n=brian@63.150.173.39) |
| 15:24.05 | *** join/#storm rafael (n=rafael@moinmoin/fan/rafael) |
| 16:12.27 | *** join/#storm jkakar (n=jkakar@72.11.116.6) |
| 16:24.09 | *** join/#storm jkakar_ (n=jkakar@72.11.116.6) |
| 17:53.25 | *** join/#storm bogdano (n=bogdano@201-15-250-176.ctame704.dsl.brasiltelecom.net.br) |
| 19:07.50 | *** join/#storm robertj (n=rcaskey@cai17.music.uga.edu) |
| 19:44.39 | *** join/#storm niemeyer (i=niemeyer@conference/europython/x-5af7cb837bc28b16) |
| 21:56.08 | *** join/#storm nephish_ (n=nephish@69.55.27.122) |
| 21:56.14 | nephish_ | hello hello ? |
| 21:58.08 | radix | nephish_: good afternoon :) |
| 21:59.32 | jkakar | Yo nephish_ |
| 22:13.51 | nephish_ | hey there, sorry, went to another desktop for a bit. I just installed storm, i am very excited about it, i use python a lot |
| 22:15.48 | nephish_ | is there any more documentation available than what is on the website tutorial ? |
| 22:17.24 | radix | nephish_: there are some docstrings |
| 22:17.38 | radix | nephish_: but nothing serious |
| 22:18.31 | nephish_ | well, an example on the tutorial shows how to create a database, but how do i connect to an existing database? |
| 22:21.24 | radix | nephish_: well, that only creates a database because sqlite in-memory databases work that way |
| 22:21.42 | radix | nephish_: you can pass a postgres:// URL to the create_database function, for example |
| 22:21.44 | radix | (or mysql://) |
| 22:22.10 | radix | "create_database" doesn't mean "create a new database object" which storm uses to communicate with your database |
| 22:23.07 | nephish_ | so create_database does not create a mysql database, but a connection to an existing database? |
| 22:23.20 | radix | right |
| 22:23.31 | nephish_ | ok, well thats easy enough. |
| 22:23.34 | nephish_ | thanks |
| 22:23.51 | radix | no problem, please ask if you have any other questions. |
| 22:25.15 | nephish_ | ok, one more. When i use mysqldb, i first had a lot of trouble with leaving connections open. We run a process here at work that parses machine reports and so on, about 16 threads, I often got crashes because of too many connection errors. |
| 22:25.41 | nephish_ | so, i wrote something that would open a connection, pass the query, return the result, and close the connection. |
| 22:26.12 | nephish_ | solved most of my woes, but still have a problem sometimes ( cannot create unix socket ) |
| 22:27.34 | nephish_ | point being, if i run several threads, connecting to one database, i thought a good ORM may help me out with some of that. |
| 22:38.11 | radix | hrm |
| 22:38.31 | radix | nephish_: I'm not sure what you're asking, exactly :) |
| 22:40.00 | nephish_ | well, i guess it's a question of how storm handles connections, is it something i need to be concerned about, or does it handle all of that by itself. |
| 22:40.03 | tjs | morning |
| 22:40.41 | radix | nephish_: there's one connection per store, meaning that Store.__init__ (basically) calls your dbapi module's "connect" function |
| 22:41.08 | radix | nephish_: you should make sure you don't use a Store from multiple threads, so if you want database access in multiple threads, create a Store per thread |
| 22:41.41 | nephish_ | so, really, it would be one connection on __init__ of my thread, then i don't have to deal with it? That would be great ! |
| 22:44.28 | radix | Ok, time to go see Harry Potter :P |
| 22:44.32 | tjs | radix: wait |
| 22:44.40 | radix | tjs: what!!! |
| 22:44.47 | tjs | radix: whasshisname said yesterday it was a db connection per thread |
| 22:44.52 | tjs | not a store |
| 22:44.55 | tjs | which one is it? |
| 22:45.03 | radix | uh, well, if you have a store per thread, then that's the same thing :) |
| 22:45.15 | tjs | no |
| 22:45.21 | tjs | create_database() per thread |
| 22:45.26 | tjs | or store |
| 22:45.44 | tjs | so you say I can create_database once |
| 22:45.49 | tjs | and that object is thread safe? |
| 22:46.09 | radix | you only need one database |
| 22:46.30 | radix | you can share the result of create_database across multiple threads, as long as you're just passing the result to the constructor of Store in each thread |
| 22:46.41 | radix | because the database object is not actually a connection, it's a factory for connections |
| 22:46.45 | radix | and Store.__init__ creates the connection |
| 22:46.46 | radix | OK bye |
| 22:46.50 | tjs | have fun |
| 22:47.26 | tjs | radix just has a thing for hermoine |
| 22:48.27 | jkakar | tjs: Heh, that was me and I implicitly meant 1-conn/store. :) |
| 22:48.36 | jkakar | Even if that wasn't what I actually said. |
| 22:49.41 | tjs | jkakar: well, I guess its not so important this morning :) I wont be doing my own storm/z3 integration anymore |
| 22:49.58 | jkakar | Cool |
| 22:50.05 | tjs | I had a chat with mark and stevea and got a more accurate idea of when zstorm will be available |
| 22:50.16 | jkakar | Oh cool. |
| 22:50.16 | tjs | so am happy to wait |
| 22:50.29 | nephish_ | zstorm? |
| 22:50.48 | jkakar | nephish_: It's our as-yet-unreleased glue to make Storm work with Zope3. |
| 22:51.03 | jkakar | nephish_: Looks like it'll be opened (at some point in the [near] future). |
| 22:51.17 | nephish_ | oh, guess i don't need that, cool |
| 22:51.40 | nephish_ | will not change the API for kind of thing i am doing, i suppose |
| 23:25.54 | *** join/#storm jkakar (n=jkakar@69-30-72-119.dq1sf.easystreet.com) |
| 23:27.47 | *** join/#storm jkakar (n=jkakar@69-30-72-119.dq1sf.easystreet.com) |