| 00:00.44 | *** join/#storm wallflower (n=wallflow@ip-205-246-113-216.pool.grokthis.net) |
| 01:21.37 | thumper | Hi guys, https://code.edge.launchpad.net/storm |
| 01:22.19 | thumper | there is now a link for active reviews |
| 01:22.35 | thumper | if you put your code reviews into "needs review" state, then you should see them there |
| 01:22.45 | thumper | also now you should be able to reply to the code review emails |
| 01:22.51 | thumper | and have them show in the web ui |
| 01:22.56 | jamesh | yay |
| 01:23.17 | jamesh | [and launchpad.net is now running on storm] |
| 01:23.53 | thumper | w00t |
| 03:50.11 | *** join/#storm nullie (n=ilya@87.254.150.23) |
| 04:27.22 | *** join/#storm jukart (n=jukart@85-124-221-45.static.xdsl-line.inode.at) |
| 05:49.05 | *** join/#storm jukart (n=jukart@lsfw01.lovelysystems.com) |
| 06:57.30 | *** join/#storm bac (n=bac@cpe-065-190-187-178.nc.res.rr.com) |
| 07:38.30 | *** join/#storm mcella (n=mcella@ip-177-245.sn2.eutelia.it) |
| 07:57.38 | *** join/#storm goschtl (n=goschtl@p5B0BEE5F.dip.t-dialin.net) |
| 08:43.19 | *** join/#storm mcella (n=mcella@ip-177-245.sn2.eutelia.it) |
| 08:55.24 | goschtl | Hi i have a class with two Unicode Attributes. If i try to str(obj.attr1 + obj.attr2) i got an unicode error. I work with german Umlaute "äüÃ". And i use sqlite as backend. any ideas? |
| 08:57.09 | therve | well it's unicode? you can't call str on it? |
| 08:58.24 | therve | maybe you want to do (attr1 + attr2).encode('your encoding') |
| 10:42.49 | *** join/#storm andrea-bs (n=andrea@ubuntu/member/beeseek.developer.andrea-bs) |
| 11:15.30 | *** join/#storm tlynn (n=tlynn@host81-149-244-57.in-addr.btopenworld.com) |
| 11:16.05 | tlynn | how does storm handle not-null constraints? |
| 11:21.29 | jamesh | tlynn: you can set allow_none=False on a column, which will prevent you from setting that column to None and raise an exception if a NULL is found in that column |
| 11:21.48 | jamesh | tlynn: but in general it is up to the DB to enforce its constraints |
| 11:21.52 | tlynn | thanks, that's exactly what I was after |
| 11:22.07 | tlynn | is it a new feature? |
| 11:22.18 | jamesh | don't think so |
| 11:22.41 | jamesh | probably about 1.5 years old |
| 11:24.21 | tlynn | found it, it just wasn't where I expected it to be in the source. thanks. |
| 11:24.46 | tlynn | probably worth putting in the docs somewhere too :-) |
| 11:26.29 | jamesh | I don't think niemeyer uses it :) |
| 11:26.42 | jamesh | I think it was added when we were working on the sqlobject compatibility layer |
| 11:32.24 | tlynn | while I'm here, is there a way to get storm to generate the create table statement? |
| 11:33.03 | tlynn | writing allow_none=True and NOT NULL for every field is going to get a bit boring :-) |
| 11:33.58 | tlynn | False even :-) |
| 11:35.03 | jamesh | tlynn: there is no support for generating the tables from Python classes in Storm |
| 11:35.47 | jamesh | tlynn: I wouldn't bother trying to replicate every database level constraint at the Python level either: it is bad from a maintenance standpoint. |
| 11:35.54 | tlynn | ok, never mind. I'll declare a NInt field-type with an allow_none=False default kwarg |
| 11:36.30 | jamesh | the DB will generally handle "NOT NULL" and column defaults for you quite well ... |
| 11:37.18 | tlynn | I agree to some degree, but I'd like to write a front-end which tells the user which field was wrong, and that seems the easiest approach |
| 11:38.12 | tlynn | I could parse the error messages, but that's even more unpleasant |
| 11:38.35 | jamesh | tlynn: generating table definitions from Python classes is something that initially sounds appealing, but has a number of problems when you put it into practice |
| 11:38.58 | jamesh | (unless you never change your application code) |
| 11:40.12 | tlynn | yep. I'm happier with the 'NInt' approach. |
| 12:05.09 | *** join/#storm niemeyer (n=niemeyer@200-138-40-171.ctame705.dsl.brasiltelecom.net.br) |
| 12:10.28 | tlynn | how do you list the fields of a storm instance? |
| 12:11.39 | tlynn | properties, I should say |
| 12:17.13 | radix | tlynn: dir() probably works |
| 12:27.39 | tlynn | well, it gets it approximately right, and with isinstance it could probably be made good enough, but the store must know the exact answer |
| 12:28.36 | tlynn | although I'd rather it turned out to be a helper function in fact, since I might want to use it on objects before they reach the store |
| 12:46.55 | *** join/#storm kov (n=kov@debian/developer/kov) |
| 13:10.07 | *** join/#storm bigdo2 (n=scmikes@72-197-8-8-arpa.cust.cinci.current.net) |
| 13:50.56 | tlynn | re. custom property types, why is there RawStrVariable.parse_set but no .parse_get? how do you go from encoded form to class? via the constructor? |
| 14:04.34 | niemeyer | tlynn: There is parse_get(), it's just unchanged from the parent class |
| 14:04.42 | niemeyer | tlynn: I don't understand the second part of the question |
| 14:07.17 | tlynn | well, if I have the data in an encoded form in the database, how do I write the code to automatically turn it back? |
| 14:08.57 | niemeyer | tlynn: Sorry, I guess I need more coffee.. I still don't get what you mean.. my brain doesn't get the meaning of "how to write the code to turn it back" |
| 14:09.11 | tlynn | back into an object |
| 14:09.44 | tlynn | class Foo() -> encoded form -> database -> encoded form ???-> class Foo(). |
| 14:10.28 | niemeyer | tlynn: Pretty much everything in Python is an object, you don't have to do anything about that.. I'm sure you know that, but it just points out why I'm having trouble trying to help you |
| 14:10.42 | niemeyer | tlynn: Ok, let me try to explain how it works |
| 14:10.50 | niemeyer | tlynn: Maybe you can find the answer to your question somewhere |
| 14:11.04 | tlynn | ok |
| 14:11.16 | tlynn | but before you do |
| 14:11.30 | tlynn | is the final step doing Foo(encoded_form) internally? |
| 14:11.43 | niemeyer | tlynn: The only thing being done in parse_set() is checking if it is a buffer(), and then converting the buffer to a string |
| 14:11.49 | niemeyer | tlynn: No matter if the value comes from the database or not |
| 14:11.58 | tlynn | ah ok, I'm confused then |
| 14:12.13 | niemeyer | tlynn: It's a responsibility of the backend to take a plain string and convert it to whatever form the database needs to get a raw string in |
| 14:12.29 | tlynn | I thought parse_set was what you override in custom properties |
| 14:12.36 | niemeyer | tlynn: You'll see that different backends will do that in different ways |
| 14:12.39 | tlynn | I'm basing it on https://lists.ubuntu.com/archives/storm/2008-March/000565.html |
| 14:12.46 | niemeyer | tlynn: It is |
| 14:13.06 | niemeyer | tlynn: Well, in custom Variables anyway |
| 14:13.22 | niemeyer | tlynn: : You can override parse_set(), parse_get(), and pretty much anything that makes sense for you |
| 14:13.50 | tlynn | ah, so there is a parse_get() method? |
| 14:13.54 | niemeyer | tlynn: Of course, dealing with private attributes is a bad idea |
| 14:13.59 | niemeyer | tlynn: Yes |
| 14:14.17 | niemeyer | tlynn: As I pointed out earlier, RawStr is just not replacing the parse_get() from the base class |
| 14:14.46 | tlynn | right, ok, I hadn't realised that was an override from Variable |
| 14:15.15 | tlynn | sorry about that, I've found the appropriate docs now |
| 14:15.44 | tlynn | what's the point of the to_db/from_db arg? |
| 14:15.47 | niemeyer | tlynn: Nice, no problem |
| 14:16.08 | niemeyer | tlynn: That says if the attribute is coming from the database, or from high-level Python code |
| 14:16.26 | tlynn | yes, but why would my code care? |
| 14:16.49 | niemeyer | tlynn: variables may chose to store a value internally in any way that makes sense |
| 14:17.15 | niemeyer | tlynn: and they may provide values to the database in a different way that they provide to user code |
| 14:17.27 | niemeyer | tlynn: Same thing about being set with database values |
| 14:17.49 | niemeyer | tlynn: E.g. it may be fine to parse a datatime string when we get one from the database, but we want proper datetime objects from user code |
| 14:17.55 | niemeyer | datetime |
| 14:18.21 | tlynn | that makes sense |
| 14:18.42 | tlynn | thanks |
| 14:18.48 | niemeyer | You're welcome |
| 14:19.05 | tlynn | while you're here, is there a way to get the list of fields of an instance other than dir()? |
| 14:19.19 | niemeyer | tlynn: Not a supported way right now |
| 14:19.46 | tlynn | ok, thanks |
| 14:26.34 | tlynn | ah, hang on. are parse_get and parse_set just buffer <-> buffer transforms? |
| 14:33.08 | tlynn | that is, what should CustomerVariable.parse_get() return? |
| 14:34.14 | niemeyer | tlynn: It should return the variable value, to user code or to the database |
| 14:34.41 | tlynn | an instance of type CustomerVariable? |
| 14:34.46 | tlynn | or a buffer? |
| 14:35.34 | tlynn | I think my problem is understanding the relative roles of Foo and FooVariable |
| 14:37.35 | niemeyer | tlynn: Variables are per instance, properties are per class |
| 14:37.47 | tlynn | let's start again. Say I have a Price value type, and an encoding of its data into a string, where does the code to convert between the two forms go |
| 14:37.50 | tlynn | (thanks) |
| 14:38.43 | niemeyer | tlynn: Either in parse_get() or parse_set(), depending on how you want to store the value internally in the variable |
| 14:38.53 | niemeyer | tlynn: Probably in parse_get() |
| 14:40.02 | tlynn | if I store it in decoded (object) form internally |
| 14:41.23 | tlynn | for ease of understanding, let's assume I'm storing JSON in the database |
| 14:42.04 | tlynn | does parse_get get called as parse_get(json_string, False)? |
| 14:42.45 | niemeyer | tlynn: s/False/True |
| 14:42.50 | niemeyer | tlynn: It's from the database |
| 14:43.06 | niemeyer | tlynn: Oh, wait.. no.. it gets called with parse_set(json_string, True) |
| 14:43.10 | tlynn | ok |
| 14:43.51 | tlynn | and that should return a Price object, or set the fields on self? |
| 14:44.13 | tlynn | (if there's documentation for this, please say so) |
| 14:46.05 | niemeyer | tlynn: It should return whatever value you want to store internally in the value |
| 14:46.11 | niemeyer | tlynn: Sorry, in the variable |
| 14:46.19 | tlynn | does it have to be compatible with set_state? |
| 14:46.33 | niemeyer | tlynn: parse_get() will be called with this value when providing the value to a user, or to the database |
| 14:46.53 | niemeyer | tlynn: In a certain way, yes |
| 14:47.07 | niemeyer | tlynn: The internal value must be comparable |
| 14:47.22 | niemeyer | tlynn: In other words, equality must make sense for it |
| 14:47.28 | tlynn | ok |
| 14:53.11 | tlynn | (reading the source) |
| 14:57.05 | tlynn | so, I've found the idea of variable factories. is "variable_class = Foo" equivalent to "variable_factory = partial(Foo)"? |
| 15:03.03 | *** join/#storm reacocard (n=reacocar@rccy-06-1010.dsl.iowatelecom.net) |
| 16:22.45 | *** join/#storm nullie (n=ilya@217.118.83.1) |
| 16:38.19 | *** part/#storm reacocard (n=reacocar@rccy-06-1010.dsl.iowatelecom.net) |
| 17:58.44 | kov | niemeyer: I made that merge we talked about, can you take a look at it when you have time? =) |
| 17:59.19 | niemeyer | kov: No doubts, thanks for pushing it |
| 18:00.31 | kov | niemeyer: let me know if I've done anything wrong when it comes to bzr handling, I am more of a svn guy, just getting to know bzr, git and other DVCS goodness |
| 18:53.44 | *** join/#storm andrea-bs (n=andrea@ubuntu/member/beeseek.developer.andrea-bs) |
| 19:01.09 | *** join/#storm lincoln (n=lincoln@201.80.63.222) |
| 19:38.02 | weatherman | store-reset r229 committed by radix@twistedmatrix.com |
| 19:38.02 | weatherman | Check for the store before deleting it from obj_info |
| 21:27.30 | *** join/#storm wallflower (n=wallflow@ip-205-246-113-216.pool.grokthis.net) [NETSPLIT VICTIM] |
| 21:27.30 | *** join/#storm radix (n=radix@wordeology.com) [NETSPLIT VICTIM] |