12:49 <djanowski> hello all. trying to use sequel with oracle XE, are there any limitations?

12:49 <djanowski> basically the SQL generated for an ALTER TABLE is different from the expected

14:05 <jeremyevans> djanowski: I've never tried. I know there are Sequel users using Oracle, but it wouldn't surprise me if not all ALTER TABLE commands work on Oracle.

14:06 <jeremyevans> djanowski: I don't have access to an Oracle installation, so I can't test it

14:06 <djanowski> jeremyevans: yeah it looks like it's not working. there's a slight difference when you add a column. nevermind, i'll test with the team and hopefully we can come up with a patch

14:06 <djanowski> jeremyevans: a good patch would mean some refactoring on alter_table_sql

14:07 <djanowski> otherwise it's hard to override for a single case

14:07 <jeremyevans> djanowski: In general, you should just have to override it in the shared Oracle adapter

14:07 <jeremyevans> djanowski: That's what the other adapters do

14:07 <djanowski> jeremyevans: override the whole method?

14:07 <jeremyevans> djanowski: The probability that it requires a change to Sequel core is unlikely

14:08 <jeremyevans> djanowski: Basically, you just override it to handle the cases that need to be handled differently, and call super for the other cases

14:08 <jeremyevans> djanowski: Look how it is overridden in the other shared adapters

14:09 <jeremyevans> djanowski: http://github.com/jeremyevans/sequel/blob/master/lib/sequel/adapters/shared/mysql.rb#L93

14:09 <djanowski> jeremyevans: still not very clean, in this case we'd only need to override one part of the `case`, not the whole ALTER TABLE statement

14:10 <jeremyevans> djanowski: Which particular case isn't working?

14:10 <djanowski> jeremyevans: ALTER TABLE "table" ADD COLUMN "foo" INTEGER

14:10 <djanowski> jeremyevans: that's what sequel is trying to execute

14:10 <jeremyevans> djanowski: What's the Oracle syntax needed?

14:10 <djanowski> but oracle doesn't want ADD COLUMN but simply ADD

14:11 <jeremyevans> djanowski: So you'd use case, override it for :add_column, and call super in the else

14:12 <jeremyevans> djanowski: If the ADD/ADD COLUMN distinction affected more than one database, I'd think about supporting it in Sequel core

14:12 <djanowski> jeremyevans: oh, sorry, no, i didn't mean that we'd need to support it

14:12 <jeremyevans> djanowski: However, if it is just for Oracle, it's better just to override the whole add_column case in the adapter

14:12 <djanowski> simply that using the case doesn't look very clean to me in terms of being able to easily override it

14:13 <djanowski> i'd be duplicating the rest of the ALTER TABLE syntax

14:13 <djanowski> s/syntax/statement

14:13 <jeremyevans> djanowski: Yes. A whole line. Wow. :p

14:13 <jeremyevans> djanowski: Really, it's not that bad

14:14 <djanowski> jeremyevans: i agree, the world has other problems to worry about :)

14:14 <djanowski> but you do agree that it's not super clean, right?

14:14 <djanowski> if that line were doing more things, for example

14:14 <jeremyevans> djanowski: Yes, it could be cleaner, but there's a tradeoff there

14:15 <jeremyevans> djanowski: If it was more than one line, I might consider a core patch, but in this case, I don't think it is worth it

14:16 <djanowski> jeremyevans: ok, no problem, it's good to know beforehand

14:16 <djanowski> jeremyevans: thanks as always

14:16 <jeremyevans> djanowski: No problem :)

15:34 <djanowski> jeremyevans: # File lib/sequel/adapters/shared/oracle.rb, line 114

15:35 <djanowski> jeremyevans: at what point does the dataset become associated with a sequence?

15:35 <djanowski> not sure how opts[:sequence] becomes true

15:38 <jeremyevans> djanowski: The sequence method (called manually), see line 129

15:39 <djanowski> jeremyevans: right, so Sequel::Model never uses it when inserting a new record?

15:41 <djanowski> jeremyevans: a bit more `ack`ing shows this: set_dataset dataset.sequence(:seq_albums_id)

15:41 <djanowski> jeremyevans: that did it

15:41 <jeremyevans> djanowski: If using Sequel::Model's with Oracle, I believe you need to set the sequence on the dataset so that the correct primary key value is returned when creating records

15:42 <jeremyevans> djanowski: This isn't required on most other databases

15:42 <djanowski> jeremyevans: is this something that could be somehow built in? maybe the adapter would be able to set the dataset?

15:42 <jeremyevans> djanowski: If there's a good way to determine the sequence for a given table automatically, I wouldn't be against doing that

15:42 <jeremyevans> djanowski: That's how the postgres adapter works.

15:43 <jeremyevans> djanowski: But I don't know how to do it on Oracle, or have any ability to test it, and nobody has contributed any code to do so

15:43 <djanowski> jeremyevans: absolutely

15:43 <jeremyevans> djanowski: So on Oracle, it needs to be set manually

15:43 <djanowski> checking the postgres adapter now

15:43 <djanowski> jeremyevans: would you be interested in getting access to an oracle instance to run all of the sequel suite on it?

15:44 <jeremyevans> djanowski: Sure

15:46 <merboutpost> jeremyevans: is there a way how to run migration to see the SQL but not execute the migration?

15:47 <jeremyevans> merboutpost: No

15:47 <jeremyevans> merboutpost: You'd have to setup a dummy database

15:47 <jeremyevans> merboutpost: Getting the SQL is easy, just add a logger

15:47 <jeremyevans> merboutpost: But making it so it doesn't execute is tricky

15:47 <merboutpost> jeremyevans: no problem, just quick reality check :-)

15:48 <jeremyevans> merboutpost: Generally, if you have a test database, just run it on that with a logger

15:48 <merboutpost> jeremyevans: yep that what i do, well I do sequel -E currently

15:49 <jeremyevans> merboutpost: sequel -E just adds a standard output logger :)

15:52 <merboutpost> I noticed that the up/down methods are actually evaluated in the different context than the migration

15:54 <jeremyevans> merboutpost: Well, yes and no

15:54 <jeremyevans> merboutpost: A migration acts as a proxy for the database

15:54 <merboutpost> jeremyevans: I defined the LANGUAGES constant, but I need to qualify it

15:54 <jeremyevans> merboutpost: So methods get send to the database using method_missing

15:55 <merboutpost> jeremyevans: Ah..

15:55 <jeremyevans> merboutpost: But constant lookup does not

15:55 <jeremyevans> merboutpost: We could add a const_missing method to work around that, I suppose

15:55 <merboutpost> jeremyevans: no neeed, I used AddMetadataToGalleriesMigration::LANGUAGES

16:41 <josh0x0> how can I get different models to use different databases? I tried setting db= ANOTHER_DB in one model, but it appears setting db = ANOTHER_DB2 in another model sets it for all models

16:45 <bougyman> yeah, DB does.

16:45 <bougyman> set the connection to a different hing.

16:45 <bougyman> DB is the default if the model isn't explicitly set to a connection.

16:45 <bougyman> so explicitly set it.

16:51 <jeremyevans> josh0x0: class Model1 < Sequel::Model(DB1); end; class Model2 < Sequel::Model(DB2); end

16:54 <jeremyevans> josh0x0: class Model1 < Sequel::Model; self.db = DB1; end; class Model2 < Sequel::Model; self.db = DB2; end # should also work

16:55 <jeremyevans> josh0x0: First method is preferred as it results in one fewer query

16:55 <josh0x0> okay. i was using the :table_name before as an argument to Sequel::Model not a Database (DB)

16:55 <josh0x0> jeremyevans: i tried the 2nd option. somewhere the DB for one of my models is being set to DB2 after I originaly set it to DB1

16:56 <jeremyevans> josh0x0: Can you provide a self contained example showing the problem?

16:56 <jeremyevans> josh0x0: Also, you can do class Model1 < Sequel::Model(DB1[:table_name]); end;

16:58 <jeremyevans> josh0x0: Also, make sure you are using self.db = and not db = inside your models

16:58 <jeremyevans> josh0x0: local variable creation instead of assignment method call and all that

16:59 <jeremyevans> Gotta go, leaving for the airport to go to MWRC

18:17 <jeremyevans> At the airport, flight leaves in 45 minutes

18:18 <jeremyevans> Sacramento's airport provides free WiFi, unlike most other airports I've been through