00:31
really_kenphused joined
02:07
<cjheath>
In documenting DataObjects base classes, RDoc finds a constant URI (assigned to be Struct.new(...)) and a class DO::URI - the same thing re-opened
02:07
<cjheath>
Is there any good reason why I shouldn't change this to be defined as just:
02:08
<cjheath>
class URI < Struct.new(:scheme, :user, :password, :host, :port, :path, :query, :fragment)
02:08
<cjheath>
Sure, there's an extra superclass, but all specs still pass, and RDoc isn't confused
02:09
<cjheath>
dbussink: ping?
02:59
beawesomeinstead joined
04:54
<manguy>
How can I do a SELECT DISTINCT with datamapper?
08:08
<botanicus>
I have problem with encoding with SQLite3, Ruby 1.9.1 and DM 0.9.11 (stable version from rubygems)
08:08
<botanicus>
http://gist.github.com/108759
08:09
<botanicus>
The data from SQLite are taken as US-ASCII, but it should be UTF-8
08:09
<botanicus>
I think it must be bug, while Encoding.default_external is set to UTF-8
08:10
<botanicus>
Any solution? Anyone has also the issue?
08:59
sporkmonger_ joined
10:50
<naeu>
I'm trying to use DataMapper in a super-simple sinatra app, however it doesn't seem to persist between calls
10:50
<naeu>
http://gist.github.com/108806
10:51
<* smoofles>
wonders why his database has over 5000 entries all of a sudden for a Model
10:52
<smoofles>
naeu: you mean the variables you fetch first seem do "forget" they are DM objects/models ?
10:53
<* smoofles>
wonders if what he asked made sense
10:53
<naeu>
smoofles: no, when i call Model#all it's empty even though I'm adding new records
10:53
<naeu>
I wonder if it might be a sinatra problem
10:54
<smoofles>
I had a problem with Model.all yesterday, when I was running it via merb -i (merb's console)
10:54
<naeu>
but i thought it might be useful to ask in here
10:54
<naeu>
the code is pretty simple (particularly the datamapper stuff)
10:54
<smoofles>
but I can't say if it has anything to do with what you're seeing
10:54
<naeu>
smoofles: does the code i've written make sense?
10:55
<smoofles>
basically I had to do "models = Model.all" every second line while in the console, or else the commands issues on the models block would have no effect :/
10:56
<smoofles>
naeu: I think it does
10:56
<smoofles>
you're reading all records when getting stuff and whatever is posted you add as a new record
10:57
<smoofles>
are you sure you want to run DataMapper.setup(:default, 'sqlite3::memory:') ?
10:57
<namelessjon>
naeu: Are you running sinatra with shotgun?
10:57
<naeu>
i'm not sure, do i?
10:57
<naeu>
namelessjon: what's shotgun?
10:58
<namelessjon>
Okay, that answers that question. :)
10:58
<namelessjon>
naeu: But as smoofles says, you probably don't want to run it with a memory db.
10:58
<namelessjon>
It will vanish when you close the app.
10:58
<smoofles>
sqlite3::memory: keeps data only while it is loaded in memory, after it is unloaded it's gone
10:59
<namelessjon>
Also, the problem line is #12
10:59
<naeu>
that's ok for this app :-)
10:59
<namelessjon>
Every time sinatra reloads in dev mode, that line gets run, wiping the DB.
10:59
<* smoofles>
missed that
10:59
<naeu>
so how do i only run the migrations once?
10:59
<namelessjon>
You want to wrap it in configure do end
11:00
<namelessjon>
Probably all the dm stuff there, from the setup call, the model and then that line.
11:00
<naeu>
namelessjon: perfect, thanks!
11:01
<naeu>
smoofles: thanks too!
11:01
<namelessjon>
naeu: But anyway, shotgun is the future of sinatra reloading (which is being removed). http://github.com/rtomayko/shotgun
11:02
<naeu>
the future that's being removed?!
11:02
<namelessjon>
reloading is being removed. shotgun will replace it.
13:13
sporkmonger_ joined
13:14
maximeguilbot joined
15:23
rubyconsumer joined
15:35
<FreakGuard_>
I've got to use DotaModel::Player.first :conditions => ["name = ? OR botuser = ?", player_name, player_name] to get a query like Fri, 08 May 2009 18:53:08 GMT ~ debug ~ (0.000035) SELECT "id", "name", "botuser" FROM "dota_model_players" WHERE name = 'b25' OR botuser = 'b25' ORDER BY "id" LIMIT 1 - I can't state a single player_name, it just queries the second one NULL
18:36
<FreakGuard_>
hm... would you consider something like that evil: http://paste.pocoo.org/show/dOJH9Ml6PAsqJoEA2UfY/ ?
18:45
<FreakGuard_>
well, except it doesn't work :/
18:46
<snusnu>
FreakGuard: i'd consider it evil if it works with super
18:46
<snusnu>
obviously you wouldn't be able to use super inside DataMapper::Resource
18:46
<snusnu>
but from inside your own class it could be possible
18:50
<snusnu>
if i haven't already pointed you to http://sick.snusnu.info/2009/04/29/extlibhook-breaks-if-hooked-method-is-redefined/ .... overwriting method (without super) could be too much for extlib
18:53
<FreakGuard_>
snusnu, so just another module and include?
18:54
<snusnu>
yep if that works (i.e. if a super call makes sense) then this is the workaround ... unfortunately that's not always desirable ... and then comes the chaining hell .. ^^
18:56
<snusnu>
this very behavior of extlib is a bug though, imo
19:05
<erwaller>
has anyone else seen strategic eager loading break when there are no instances of the child model for a given instance of the parent model? IE, SEL works while iterating over posts with comments, but posts with no comments cause dm to issue new queries
19:06
<snusnu>
erwaller: are you on dm-core/next ?
19:08
<snusnu>
erwaller: SEL is broken in master, and dkubb currently is adding it back to the next branch ... i don't even think it's in there completely yet ... but i heard him say that 0.10.0 (the next branch) will be available soonish
19:09
<erwaller>
where can I grab that? I've tried it on dkubb's branch with no luck
19:10
<snusnu>
http://github.com/datamapper/dm-core/tree/next
19:13
<snusnu>
erwaller: i heard dkubb say that it will be released in the next couple of weeks
19:14
<erwaller>
yeah still broken there.. sounds good, as long as it's a known issue and not me doing something wrong haha
19:14
<erwaller>
thanks for the help
19:14
<snusnu>
erwaller: tell dkubb what's broken, now that he's here :)
19:17
<erwaller>
dkubb: I just ran into a problem with SEL and wanted to make sure you heard about it
19:17
<erwaller>
essentially, SEL works while iterating over posts with comments (for example), but posts with no comments cause dm to issue new queries
19:17
<erwaller>
snusnu: good call
21:37
<snusnu>
http://github.com/bmabey/database_cleaner/tree/master now has support for datamapper transaction and truncation strategies ... no more excuses for not using datamapper in a cucumber/webrat/selenium setup (or any other setup where transactional fixtures are not possible)
21:40
<cjheath>
snusnu: Who receives github pull requests sent to datamapper"?
21:40
<snusnu>
cjheath: i'd say dkubb at least
21:42
<cjheath>
mmmmok. He seemed pleased that I was adding doco to the DataObjects base API, but no-one's responded to the pull req. Guess RC is taking a tol ;-)
21:43
<snusnu>
cjheath: for DO i think dbussink is the man
21:43
<cjheath>
I pinged him here, but he's cnnected more often than he's present ;-)
21:44
<snusnu>
cjheath: yeah, assuming RC is railsconf ... he was there :) ... and dbussink ... i *think* i heard him say that he'll be at euruko this weekend
21:45
<cjheath>
np anyhow
21:45
<cjheath>
Just making a start on an sqlserver adapter
21:47
<snusnu>
cjheath: nice! ... concerning adapters ... i think rando is the man behind the adapter api ... you should maybe checkout his fork too ... not too sure about how/when/if his stuff gets merged in though ... but it probably will sooner or later ...
21:47
<cjheath>
Ok, thanks for the heads-up
21:48
<cjheath>
I will need reflection, so I fgure I'll become intimate with the APIs, which is the main reason for building an adapter
21:49
<snusnu>
cjheath: http://github.com/paul/dm-core/commits/next is rando's fork
21:50
<cjheath>
Miine's here: http://github.com/cjheath/do/tree/next
21:50
<snusnu>
cjheath: ah so you're the man behind polyglot and treetop :)
21:51
<* snusnu>
had lots of fun with treetop a year ago
21:51
<cjheath>
not treetop, just a mintainer
21:51
<cjheath>
Nathan Sobo is the "Absentee Creator" ;-)
21:51
<snusnu>
hehe yeah .. haven't read anything from him in ages :)
21:52
<cjheath>
I adopted it because no traditional parser generator can parse my CQL language
21:52
<cjheath>
which is part of the activefacts gem - that's what I'll be using DO with
21:52
<snusnu>
cjheath: i quickly read through your page on activefacts .. sounds interesting!
21:53
<cjheath>
I'm working on a 10-year goal here ... replace SQL ;-)
21:53
<cjheath>
amongst other outcomes...
21:54
<snusnu>
hehe ... great goals are nice to have! :)
21:56
<cjheath>
We (the software industry) doa terrible job of comminucating with "the business", so this addresses that angle too
21:57
<dkubb>
cjheath: sorry I haven't responded during RC. lots of talking, but not as much hacking
21:58
<cjheath>
sure, np.
21:58
<snusnu>
cjheath: yeah! and not only *business* wins .. also *we* the software devs .. i as a programmer find your approach easier to read than sql as well :)
21:59
<cjheath>
I hope so. It's structures, but meant to read like plain english (or other languages). Did you look at the "Queries" section inside my OSDC presentation slides?
21:59
<cjheath>
*structured*
22:00
<cjheath>
It also includes automatic units conversions
22:01
<cjheath>
which is pretty cool when you see it.
22:02
<dkubb>
in general when you need something merged, it's best to contact the gem maintainer: http://datamapper.org/doku.php?id=gem_maintainers
22:03
<cjheath>
Ahhh, thanks, I hadn't seen that list
22:03
<snusnu>
cjheath: to be honest i haven't read all in detail yet, only skimmed through it a few days ago when i saw you posting a link to it (here i think) ... i will definitely have a closer look though1
22:03
<dkubb>
I wish there was a way to set up an "auto-responder" for some aliases, like the datamapper github user, which I don't check as often as my "dkubb" account (which is daily, sometimes hourly with how much I use github)
22:03
<cjheath>
You could set up to forward it to your email though
22:04
<dkubb>
cjheath: did you say you were thinking about doing a DO sqlserver driver?
22:05
<cjheath>
Well, I helped kick off the current round of integration of the *many* sqlserver AR adapters, and now that's stabilised, it shouldn't be hard to leverage
22:06
<dkubb>
cjheath: I ask because I met someone at RailsConf who was interested in helping on one, but he doesn't know C (but I assume would help with specs/docs). there was also someone a couple of days ago on lighthouse asking to help with it
22:06
<dkubb>
cjheath: that's really cool. were you at RailsConf this year?
22:06
<cjheath>
No need for C (Not this year)
22:06
<dkubb>
cjheath: how many sqlserver AR adapters are there?
22:06
<cjheath>
The ODBC and ADO adapters do it just fine
22:07
<cjheath>
only one now ;-)
22:07
<cjheath>
http://github.com/rails-sqlserver
22:08
<cjheath>
We merged about 12 different ones
22:08
<cjheath>
I did the main merge of 8
22:08
<cjheath>
and then others fired up and made everything work really nicely
22:08
<dkubb>
cjheath: ahh ok. I'm fine with that in the interim. ideally what we'd like is 3 types for each driver: JDBC based, MRI (or FFI) C based, and pure Ruby
22:08
<cjheath>
Can't do pure Ruby
22:08
<dkubb>
cjheath: there's some interest in having an AR DO adapter written
22:09
<cjheath>
unless you want to recode the freetds library in Ruby - big job!
22:09
<dkubb>
cjheath: would you be basing it off the freetds lib?
22:09
<cjheath>
The wire protocol has been almost completely rev-eng'd
22:10
<cjheath>
the ODBC adapter is based on freetds
22:10
<cjheath>
AR builds on that
22:10
<cjheath>
too many layers,
22:10
<dkubb>
cjheath: (sorry if I bombard you with questions, this is something I'm a little uneducated on, and I'd like to understand so I can explain it to others)
22:10
<cjheath>
I tried to do FFI/freetds
22:10
<cjheath>
but FFI was too young
22:10
<cjheath>
reetds has some funky constructs
22:10
<cjheath>
it might be feasible now
22:11
<cjheath>
butt you still want to use ADO for Windows
22:11
<dkubb>
cjheath: well, I believe: 1) make it work, 2) make it right, 3) make it fast. we need #1 first ;)
22:11
<dkubb>
cjheath: so freetds is cross platform compatible?
22:11
<cjheath>
Doing an AR meta-adapter is an interesting idea
22:11
<cjheath>
Yes, but you wouldn't use it on Windows
22:12
<cjheath>
where the authentic thing is available
22:13
<cjheath>
I don't like some of what ODBC does with data type conversions, it hides/changes some of the details
22:14
<dkubb>
cjheath: one of the things we like to do with DO is make it so installation is as simple as possible. for example, we won't have a separate do_mysql-jruby or do_mysql-mri package. you just install do_mysql and it does the right thing on your platform
22:14
<cjheath>
So when does rando's dm-more stuff (moving adapters) go live?
22:14
<dkubb>
cjheath: with different versions of sqlserver, do you think this is feasible
22:14
<cjheath>
Sure, that'd be nice
22:14
<cjheath>
freetds is a bit of a pain
22:14
<dkubb>
cjheath: when dm-core 0.10.0 and dm-more passes all the specs with it. probably within a week or so
22:15
<cjheath>
Well, the AR adapter already needs to detect which version is has
22:15
<cjheath>
2000 differs from 2005 from 2008
22:15
<cjheath>
date/time type handling in particular
22:17
<cjheath>
Without FFI/freetds, you could still do an ext/freetds thing
22:17
<dkubb>
cjheath: on the DDL/DML SQL generation side, are there large differences between 2000, 2005 and 2008?
22:18
<cjheath>
No, just type support, at least as far as AR is concerned
22:18
<dkubb>
cjheath: one of the approaches we take is all the SQL generation is done in the DataObjectsAdapter, and we output ANSI 92 compatible DDL/DML. However, each adapter specific to a DB inherits from it, and can override anything it needs to do differently
22:19
<dkubb>
cjheath: just so I'm clear, by type support, do you mean the process of coercing the value into a Ruby object based on the column type?
22:19
<cjheath>
2000 has no Date type
22:19
<cjheath>
only datetime
22:19
<cjheath>
so to store and reflect a date, you get f-ups with timezones etc
22:20
<cjheath>
The SQL generation model sounds fair
22:21
<cjheath>
I think DM knows what type to use, it doesn't use reflection and mapping, yes?
22:23
beawesomeinstead joined
22:23
<dkubb>
cjheath: how it works with DM, is that if it sends over an SELECT statement, it explicitly specifies each column, and the Ruby type it expects. the DO driver is responsible for coercing the value it gets into that type directly and returning it
22:24
<cjheath>
Right, nice, I have that in AF too
22:24
<cjheath>
How does DO know how to make an instance of the type? Is there a standard interface?
22:24
<cjheath>
Because in AF, it's pretty special
22:25
<cjheath>
you don't instantiate a class, you *assert* the existence of an instance in the current constellation
22:26
<dkubb>
cjheath: DO is also meant to work without an explicit type for each column, in which case it just does it's best
22:26
<cjheath>
"constellation" = extended result set, possibly result of more than one query
22:30
<dkubb>
cjheath: Brian Hogan (bphogan) said he really wants to help with a DO sqlserver driver, and there was a ticket added a few days ago about it http://datamapper.lighthouseapp.com/projects/20609/tickets/841-add-support-for-microsoft-sql-server-eg-do_sqlserver#ticket-841-4
22:36
<cjheath>
Thanks, I'll call him
22:36
<dkubb>
cjheath: do you know Brian?
22:36
<cjheath>
No, I don't think so
22:36
<dkubb>
cjheath: oh, I assumed you did when you said you'd call him :)
22:36
<cjheath>
Which Brian?
22:36
<cjheath>
No, I mean I'll email himm
22:36
<dkubb>
cjheath: he just did a presentation at RailsConf about making Rails work with legacy DBs. DM's a better fit for that than AR due to the ability to remap/alias columns, etc. although there wasn't enough time before his presentation to get an sqlserver driver working and tested
22:36
<dkubb>
cjheath: DM works with Rails now, but it's not seamless. I'm going to be working on that shortly
22:36
<cjheath>
I ran the "Enterprise Rails" BOF at RC 2 years ago
22:36
<cjheath>
more than 100 folk attended
22:36
<cjheath>
lots of enthusiasm
22:36
<dkubb>
by seamless I mean: "takes less than 10 minutes to setup"
22:36
<cjheath>
but enterprise folk are almost never able to contribute
22:36
<cjheath>
I still have my notes, and none of the ten prolem areas have been fixed
22:36
<cjheath>
almost all to do with DBMS
22:36
<dkubb>
I think i'll be able to make it so using DM in AR is as simple as commenting out a few AR specific configs, and installing a plugin
22:36
<dkubb>
cjheath: would you mind forwarding me those notes? (dan.kubb [at] gmail [dot] com)
22:36
<cjheath>
Will dig them up
22:36
<dkubb>
cjheath: there was a *ton* of interest in DataMapper at RailsConf
22:36
<cjheath>
Got a lot of resonance
22:36
<cjheath>
Esp now that MERB and Rails look to be merging
22:37
<cjheath>
I enev have audio of the BOF still
22:37
<dkubb>
my goal is to have it so that by this time next year somewhere between 1/10th to 1/4th of new Rails apps use DM instead of AR
22:38
<dkubb>
I think that'd be pretty awesome
22:38
<cjheath>
The audio is 46MB
22:38
<dkubb>
cjheath: sweet. if you don't mind emailing me the notes/audio I'd appreciate it.. or I could download it from somewhere
22:40
<cjheath>
It'll be slow (cable modem) but here: http://cliffordheath.homedns.org:28080/tmp/
22:41
<cjheath>
Some details are out of date now, of course
22:42
<dkubb>
cjheath: oh, that's cool. I started doing AR/Rails stuff in 2005, so I should be able to sort that out from the stuff that's current
22:43
<dkubb>
cjheath: I'
22:43
<dkubb>
whooops. I'll have the mp3 in about 21 mins or so
22:44
<cjheath>
Saturday AM here, you might get 60KB/s or better
22:45
<dkubb>
cjheath: oh, hmm. I'm getting 28 KB/s right now. my connection maxes out at about 300 KB/s
22:46
<cjheath>
Right, it jumps from 15 to 30 to 60 at most
22:46
<dkubb>
cjheath: np. I'm about to watch Slumdog Millionaire anyway, so when I come back to it, it'll be long finished
22:46
<cjheath>
hehe, cool, enjoy
22:49
<dkubb>
cjheath: I have an open offer to any potential DO authors.. if you write the DO side, and get it passing the shared specs, I'll write the whole DataMapper adapter, including full specs and docs
22:49
<cjheath>
cool, thx
22:49
<dkubb>
cjheath: you're welcome to do the DM side too, but I figure it'll help people get up and running faster. I know DM pretty well, and you as an author know your DB better than I do
22:50
<cjheath>
I don't really need to know DM yet... I'm using DO to build AF ;-)
22:50
<dkubb>
cjheath: heh
22:51
<dkubb>
cjheath: well, at the very least having a DM adapter for it will get some more real world testing, which helps the quality of the DO drivers (or so we've found). the more libs that support DO the better the quality gets
22:52
<cjheath>
Sure, and that'd be a good thinbg
22:52
<dkubb>
that's why I'm pushing so hard to get an AR DO adapter
22:52
<dkubb>
Sequel already supports DO
22:52
<cjheath>
I just have to conserve brainspace
22:52
<dkubb>
cjheath: that's understandable. building AF is a tough project
22:52
<cjheath>
It's a hydra...
22:53
<cjheath>
The basic runtime API isn't too big actually
22:53
<cjheath>
I've already spent the equivalent of 12 months fulltime (spread over 2 years) on it
22:53
<cjheath>
the whole project I mean
22:56
postmodern|cafe joined
22:57
<postmodern|cafe>
when does :default option take effect for properties?
22:57
<dkubb>
cjheath: yeah, I think it's important that there be alot of different projects trying to solve these problems from different angles
22:57
<postmodern|cafe>
should it apply for Model.new
22:58
<dkubb>
postmodern|cafe: two cases: when you ask for the value of the attribute in a new record, or when persisting a new record for the first time
22:58
<cjheath>
BTW, some of the "key advantages" of DM over AR as listed on the project home page... are now in Rails also... might need to revise a bit
22:58
<dkubb>
cjheath: oh, there's a ton of things we don't advertise yet on the home page
22:59
<cjheath>
I'm sure that;s true too... but things like the way associations get loaded for eg
22:59
<dkubb>
cjheath: plus some of those things in AR only are in very specific cases, but without an IdentityMap, there are still alot of cases where it doesn't match the behavior we have in DM. not saying DM is better than AR, but there are still a ton of differences
23:00
<dkubb>
parent.children.first.parent.should equal(parent) is one case now in AR due to their inverted relationships stuff
23:00
<cjheath>
an AF constellation does identity mapping for all values, not just rows
23:00
<cjheath>
and you only ever declare a relationship from one end
23:00
<cjheath>
There is no has_many - it gets created by has_oe
23:01
<cjheath>
You mentioned Brian Hoogan... but the Lighthouse ticket is from Greg Houston... confusion?
23:02
<dkubb>
cjheath: no, I meant to say earlier that both those guys want to help
23:02
<dkubb>
cjheath: sorry I wasn't clear
23:02
<dkubb>
cjheath: there's alot of interest from people using the "enterprise" DBs to have DO drivers.. there's an Oracle one starting up now, and possibly a DB2 one too
23:03
<dkubb>
and now sqlserver. which is awesome
23:03
<dkubb>
push that complexity into a consistent API, and you now make writing the ORM side of things alot simpler
23:03
<cjheath>
google search bphogan... "Did you mean bogan" ;-) :-)
23:04
<dkubb>
http://twitter.com/bphogan
23:04
<dkubb>
cjheath: ok. my wife is telling me to close my laptop now and watch the movie. I'll catch up with you a bit later if you're on
23:04
<cjheath>
Yep, got that