AR adaptor: forces non-numeric attributes to be 0

[yob]

I'm assuming you're planning for the mysql2 AR adaptor to be a drop in replacement for the standard mysql one? Let me know if that isn't the case and I'll stop filing these bugs.

In the standard mysql adaptor if I have a model with a decimal attribute and validates_numericality_of, trying to set that attribute to something silly like an array causes a validation error.

class Supplier < ActiveRecord::Base
  ...
  validates_numericality_of :margin, :greater_than_or_equal_to => 0, :less_than_or_equal_to => 100, :allow_blank => true
  ...
end

>> Supplier.new(:margin => []).valid?
=> false
>> Supplier.new(:margin => 0).valid?
=> true

With the mysql2 adaptor, using an array is accepted:

>> Supplier.new(:margin => []).valid?
=> true
>> Supplier.new(:margin => 0).valid?
=> true

[brianmario]

Yeah I'm definitely shooting for a drop-in replacement, or as close to it as I can get ;) No worries on reporting these issues, keep 'em comin!

I'll try and take a look at this one soon

[brianmario]

Ok try it now (http://github.com/brianmario/mysql2/commit/e01c7581e5a34fc25920b487d8a2ad102dd8b2d1)

[yob]

That commit causes any test in my app that tries to save an AR object to fail :)

All models that have a related object (Invoice belongs_to Company, a valid invoice instance fails to save because it thinks Invoice#company is nil).

[brianmario]

Any other info about why they aren't saving?

[brianmario]

Btw, I'm running my adapter against the rails test suite for the existing mysql gem and adapter to try and get those all passing

[yob]

OK, that was a really lame comment with no details. My aplogies :)

I have a company model that has the following validation:

class Company < ActiveRecord::Base
  ..
  validates_numericality_of :standard_float, :greater_than => 0, :allow_nil => true
  ..
end

The company model is created in nearly every test, so a failure in saving it triggers a sea of red across my test suite.

>> c = Company.new
=> #<Company id: nil, standard_float: nil>
>> c.save!
ActiveRecord::RecordInvalid: Validation failed: Standard float must be greater than 0

So it's calling c.standard_float invalid, even though my validation rule says nil is ok.

[brianmario]

Ok, hopefully this is the last iteration on this casting madness :P http://github.com/brianmario/mysql2/commit/6b285dcd85698e427d0498c6855f831572b6747a

[brianmario]

Whoops, pushed a typo in that commit, it's been fixed so try with the latest

[yob]

Down from 11 failures to 7! Running late for a dinner thing, so will send you detailed into later tonight.

[yob]

(and the failures look like a different issue)

[yob]

>> Supplier.last.update_attributes!(:payment_days => [])

NameError: undefined local variable or method value' for #<Supplier:0x7ff10c474a50> from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/attribute_methods.rb:260:inmethod_missing' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/attribute_methods.rb:211:in payment_days' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:466:insend' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:466 from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:465:in each' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:465 from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:182:incall' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:182:in evaluate_method' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:166:incall' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:90:in run' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:90:ineach' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:90:in send' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:90:inrun' from /var/lib/gems/1.8/gems/activesupport-2.3.5/lib/active_support/callbacks.rb:276:in run_callbacks' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:1098:invalid_without_callbacks?' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/callbacks.rb:315:in valid?' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/validations.rb:1087:insave_without_dirty!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/dirty.rb:87:in save_without_transactions!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/transactions.rb:200:insave!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/connection_adapters/abstract/database_statements.rb:136:in transaction' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/transactions.rb:182:intransaction' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/transactions.rb:200:in save!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/transactions.rb:208:inrollback_active_record_state!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/transactions.rb:200:in save!' from /var/lib/gems/1.8/gems/activerecord-2.3.5/lib/active_record/base.rb:2635:inupdate_attributes!' from (irb):4>>

[yob]

I should add that Supplier.payment_days is an int attribute defined like so:

`payment_days` int(11) NOT NULL DEFAULT '30',

with the following validation on it:

validates_numericality_of :payment_days, :greater_than_or_equal_to => 0

[brianmario]

Damn, had another typo... pushed - try one last time? (sorry if this is getting annoying :P)

[brianmario]

I thought my tests covered these cases, but apparently not... Will add more soon

[yob]

With 86b9ffe I drop from 7 failures to 4, but with 48db89 it jumps back up to 154.

The 4 failures are on methods where I'm doing a direct call to SomeModel.connection.select_all(....). All values returned are strings, so I'm manually casting the decimals to BigDecimals. That fails with mysql2, as they're already decimals. Arguably I'd say you could leave that as is, people shouldn't be dropping down to raw SQL too often.

Most of the 154 failures seems to be "String can't be coerced into Fixnum" and are triggered by aggregate functions on associations returning a string instead of a number:

>> CustomerOrderItem.last.approval_items.sum(:quantity)
=> "0"

[brianmario]

Was afraid of that, I reverted 48db89. Really sucks that I have to leave this casting stuff in because AR makes incorrect assumptions about the data all over in places I don't have control over :( Anyway, if things after this revert go back to "normal" and have only the 4 failing tests; I may push 0.1.6 - as for those tests, could just check if it's already a BigDecimal before attempting to cast and have them all pass?

[yob]

OK, b676a3 takes me back to 4 failures.

I've adapted my model code that is calling select_all() to prevent it passing a BigDecimal back into the BigDecimal constructor, so now my full suite is running green.

Thanks!

[brianmario]

awesome, thank you soooo much for all the testing!