after_initialize/after_find misfeature
Piers Cawley
object is instantiated from the database, but I'm equally sure that
it's a bad idea - if nothing else, initialize doesn't get called
anywhere in the instantiate method chain.
More importantly though, it's a huge performance hit for any
programmer who wants to do something like setting defaults on their
object before it gets written to the database, or even checked for
validity. At present, the only way to get callback behaviour that
triggers only when an object is first made is to do something like:
def after_initialize
if new_record?
do_stuff
end
end
But now she is paying the cost of a method call not only when an
object is initialized (which is fine, because the method actually
_does_) something, but also every time objects of that class are
pulled from the database (an eventuality which is already handled by
the after_find callback.
Being strict and only calling after_initialize from
initialize_with_callbacks and leaving instantiate_with_callbacks only
calling after_find can only make after_initialize more useful. If
someone really does need behaviour common to both points in an
object's lifecycle, it's easy enough to do:
after_find :common_behaviour
after_initialize :common_behaviour
and god is in his heaven and all is right with the world.
If the cost of changing the API in this way is deemed too high, please
at least consider adding some callback that really is only called
after initialization and not after instantiation. It's too useful a
place to ignore.
Michael Koziarski
> programmer who wants to do something like setting defaults on their
> object before it gets written to the database, or even checked for
> validity.
We have before_validation for this case.
> If the cost of changing the API in this way is deemed too high, please
> at least consider adding some callback that really is only called
> after initialization and not after instantiation. It's too useful a
> place to ignore.
Taking the case of
1: @customer = Customer.new(params[:customer])
2: @customer.do_something
3: @customer.save!
What we currently don't have is a hook that will fire before line 2,
but not when retrieving from a database. Providing defaults before
validation was a common case, and I believe that's where the
before_validation hook came from. What other use cases do you have
in mind?
The name after_initialize is probably a little misleading because it
fires even when an object is retrieved from the database. Strictly
speaking this is 'initialization', but from a user's perspective it's
'finding' or 'retrieval'. Perhaps at the very least we could give it
a less confusing, more hazardous sounding name.
--
Cheers
Koz
Piers Cawley
>
> > More importantly though, it's a huge performance hit for any
> > programmer who wants to do something like setting defaults on their
> > object before it gets written to the database, or even checked for
> > validity.
>
> We have before_validation for this case.
No, you don't.
class Foo < ActiveRecord::Base
def before_validate
self.attrib_with_default ||= 99
end
end
p Foo.new.attrib_with_default # nil
The default doesn't get set until the object is about to be validated,
but sometimes you need that default to be set correctly so that you
can use an object before it's either validated or saved. And are you
really suggesting that everyone should call 'valid?' on their objects
if they want them to get their correct defaults set? Default setting
belongs in the initialization phase, not in the validation phase.
> > If the cost of changing the API in this way is deemed too high, please
> > at least consider adding some callback that really is only called
> > after initialization and not after instantiation. It's too useful a
> > place to ignore.
>
> Taking the case of
>
> 1: @customer = Customer.new(params[:customer])
> 2: @customer.do_something
> 3: @customer.save!
>
> What we currently don't have is a hook that will fire before line 2,
> but not when retrieving from a database. Providing defaults before
> validation was a common case, and I believe that's where the
> before_validation hook came from. What other use cases do you have
> in mind?
Well, the obvious case is the case where you're implementing new in
your controller.
The typical boiler plate goes:
def new
@customer = Customer.new
end
That @customer is never going to be validated, but it does make sense
for it to have its defaults (which belong in the model and not the
controller or the view) set correctly so that 'new.rhtml' doesn't need
any knowledge of any default values when rendering the form.
> The name after_initialize is probably a little misleading because it
> fires even when an object is retrieved from the database. Strictly
> speaking this is 'initialization', but from a user's perspective it's
> 'finding' or 'retrieval'. Perhaps at the very least we could give it
> a less confusing, more hazardous sounding name.
But 'after_initialize' implies "call this after you call the
instance's initialize method", and it's apparent that, in the case of
object retrieval, the initialize method simply isn't called (which
makes sense, because the object is only initialized once, when it got
created by calling its class's new method.)
Damian Janowski
> def new
> @customer = Customer.new
> end
>
> That @customer is never going to be validated, but it does make sense
> for it to have its defaults (which belong in the model and not the
> controller or the view) set correctly so that 'new.rhtml' doesn't need
> any knowledge of any default values when rendering the form.
Really hear you on that one.
If we need to deal with default values in a better way, why not
something like...
class Foo < ActiveRecord::Base
default :attrib, :to => 1
default :another_attrib, :to => :another_attrib_defaulter
protected
def another_attrib_defaulter
# ...
end
end
The second parameter being a hash for some readability and potentially
for some more features...
Thoughts?
court3nay
On Jul 22, 2007, at 1:14 PM, "Damian Janowski" <damian....@gmail.com
> wrote:
Does default belong in the schema as it is essentially a db
functionality? Or is there a case for more complex conditional logic?
In the latter case, what about using an overloaded attr reader instead?
courtenay
Piers Cawley
No it isn't. Consider the use case of the newly created 'template'
object that's used by new.rhtml in some generic controller; this never
goes near the database, but still needs to have its defaults correctly
set.
Michael Koziarski
> but sometimes you need that default to be set correctly so that you
> can use an object before it's either validated or saved. And are you
> really suggesting that everyone should call 'valid?' on their objects
> if they want them to get their correct defaults set? Default setting
> belongs in the initialization phase, not in the validation phase.
You'll note that this is line two mentioned below. Your original
email cited "before checked for validity", and before_validation does
that. Either way, we're talking past one another here.
> But 'after_initialize' implies "call this after you call the
> instance's initialize method", and it's apparent that, in the case of
> object retrieval, the initialize method simply isn't called (which
> makes sense, because the object is only initialized once, when it got
> created by calling its class's new method.)
This thin and literal interpretation of the word 'initialize' makes
sense only to those who care deeply about the implementation of AR.
If we were to refactor Base.instantiate(record) to use .new, I'm not
sure that many users would care, nor would they expect different
callbacks to be called.
I'm not rejecting the need for defaults beyond what the schema
provides, merely suggesting that there's probably a better name out
there for the callbacks to achieve that behaviour.
--
Cheers
Koz
court3nay
wrote:
Right, that's the point i was missing.. but couldn't you do that as an
overloaded reader?
def name
read_attribute[:name] || "default name"
end
I do like the idea of wrapping that functionality in your syntax as
above, makes it a little quicker
courtenay
> >
Michael Koziarski
you can use this in some situations, but if you have interdependent
defaults for three or four fields it can get really confusing. Better
to have that logic put in one common place, and a callback is probably
the right place for it.
--
Cheers
Koz
Piers Cawley
> > instance's initialize method", and it's apparent that, in the case of
> > object retrieval, the initialize method simply isn't called (which
> > makes sense, because the object is only initialized once, when it got
> > created by calling its class's new method.)
>
> This thin and literal interpretation of the word 'initialize' makes
> sense only to those who care deeply about the implementation of AR.
> If we were to refactor Base.instantiate(record) to use .new, I'm not
> sure that many users would care, nor would they expect different
> callbacks to be called.
I really do have to take issue with this; I would argue that the
current behaviour of after_initialize is based on a 'thin and literal
interpretation of the word "initialize"'. Decanting an object from
storage is not initialization; initialization is what happens when an
object is first created, back before it got poured into the storage
jar.
Consider a woollen gansey's life cycle. It starts off as a length of
wool which is then knitted (initialized) on circular needles into a
delightful seemless garment. During cold weather, I wear it a lot.
During warm weather, I put it in a drawer (storage). When it gets cold
again, I don't take the gansey out of the drawer and knit it again
from scratch.
Conceptually, that's what's happening with my active record object.
It's knitted, shoved into the drawer and pulled out again as needed.
It only gets knitted once. Of course, what _really_ happens is that my
'gansey' is stored in the teleporter's pattern buffers (okay, if you
insist, we'll call it a database) and reconstituted from an entirely
new set of atoms when I need it later. But that doesn't matter. As far
as the wearer is concerned, it's the same gansey.
Plus, there's the argument from orthogonality... If after_initialize
only fires when the gansey is first knitted, then I can add behaviour
that fires after knitting and drawer removal by doing:
after_find :do_something
after_initialize :do_something
But when after_initialize fires after find as well, I have to write
after_initialize do |obj|
if obj.new_record?
...
end
end
and pay the cost of the method call every time I get a gansey out of
the drawer anyway. No fun.
Michael Koziarski
> current behaviour of after_initialize is based on a 'thin and literal
> interpretation of the word "initialize"'. Decanting an object from
> storage is not initialization; initialization is what happens when an
> object is first created, back before it got poured into the storage
> jar.
I can buy your argument, but if I do so, there doesn't seem to be a
need for after_initialize:
def initialize(attrs)
do_stuff
super
end
All we'd be doing is reinventing an initialize method which hid the
arguments. Doesn't seem particularly necessary, especially at the
cost of backwards compatibility for the people who use
after_initialize at present.
--
Cheers
Koz
Piers Cawley
Have you actually tried that? I have. It's a nightmare. It's long
enough ago now that I can't remember exactly what the nightmare was,
but it was definitely no fun at all.
A cursory examination ActiveRecord::Base suggests that it's because
the various attributes don't work until after '@attributes =
attributes_from_column_definition' has been evaluated, and that gets
evaluated when you call @super.
I accept that there are people who rely on the current behaviour of
after_initialize, so the trick will probably be to come up with a good
name for the callback. 'after_new'?
Another option might be to turn ActiveRecord::Base#initialize into a
template method along the lines of:
def initialize(attributes = nil)
@attributes = attributes_from_column_definition
@new_record = true
ensure_proper_type
initialize_defaults if self.respond_to?(:initialize_defaults)
self.attributes = attributes unless attributes.nil?
yield self if block_given?
end
which has the advantage of making the callback after the bones of the
object are in place, but before any @attributes are assigned to.
Jonathan Viney
http://svn.viney.net.nz/things/rails/plugins/active_record_defaults/
-Jonathan.
Tom Ward
> On 22/07/07, Michael Koziarski <mic...@koziarski.com> wrote:
> > def initialize(attrs)
> > do_stuff
> > super
> > end
> Have you actually tried that? I have. It's a nightmare. It's long
> enough ago now that I can't remember exactly what the nightmare was,
> but it was definitely no fun at all.
>
> A cursory examination ActiveRecord::Base suggests that it's because
> the various attributes don't work until after '@attributes =
> attributes_from_column_definition' has been evaluated, and that gets
> evaluated when you call @super.
If you want the equivalent of after_initialize, wouldn't you want to
do_stuff after calling super? In any case, if you're feeling funky,
you can do:
def initialize(attrs = {}, &block)
attrs[:funky_default] ||= 'my funky default'
super
end
But I much prefer:
def initialize(attrs = {}, &block)
super
self.funky_default ||= 'my funky default'
end
Tom
Michael Koziarski
> do_stuff after calling super? In any case, if you're feeling funky,
> you can do:
>
> def initialize(attrs = {}, &block)
> attrs[:funky_default] ||= 'my funky default'
> super
> end
>
> But I much prefer:
>
> def initialize(attrs = {}, &block)
> super
> self.funky_default ||= 'my funky default'
> end
Sorry, that's what I meant, if you do stuff before @attributes is
initialized you'll get some nasty surprises. Without cases which
can't be solved by this pattern, I'm not sure that adding a new
callback is justified.
--
Cheers
Koz
Ben Munat
Sorry but I must respectfully disagree. This issue caused me plenty of confusion when I started out
with rails, I've seen lots of other people being confused about it, and you even admit here that if
you don't do it right you'll get "nasty surprises". Any good framework should try to avoid the
potential for nasty surprises where possible.
Now, Jonathan's plugin mentioned earlier in this thread sounds like a pretty good solution, but I
had never heard of it until now. Maybe core could consider adopting that. But, I would say that
setting default values for model objects is an *extremely* common practice... needed on just about
every project. It seems like one way or another it would be nice to help rails developers avoid some
gotchas with something that's completely straightforward in most other languages/frameworks.
Ben
Pratik
> needed on just about every project.
Just wondering how did you achieve it previous to hearing about
Jonathan's plugin ?
Why add a new callback when all you need is 4 lines of documentation
on how to do it ?
--
Cheers!
- Pratik
http://m.onkey.org
Ben Munat
>> But, I would say that setting default values for model objects is an *extremely* common practice...
>> needed on just about every project.
>
> Just wondering how did you achieve it previous to hearing about
> Jonathan's plugin ?
>
> Why add a new callback when all you need is 4 lines of documentation
> on how to do it ?
>
a while back where he outlined all the gotchas with setting default values. It was actually my
impression from that post that one could *not* override initialize.
So, my point really is that needing to override after_initialize, but remember to put the check for
a new_record in there (plus that fact that it's run on every load) is silly... and
counter-intuitive. But, if overriding initialize is really completely safe and won't cause lots of
heartache then fine. I'm not one to stampede to adding stuff to the framework.
I guess the question is, how bad a deal is it for people who miss the four lines of documentation
versus how much of a change is to to provide a simple, obvious way to set model defaults. This is
just one of those things that I need to explain to rails newcomers that seems silly and a bit
embarrassing.
Anyway, I think I'll definitely check out Jonathan's plugin on my next app... looks all nice and
"DSL-ish". :-)
Ben
Piers Cawley
> Anyway, I think I'll definitely check out Jonathan's plugin on my next app... looks all nice and
> "DSL-ish". :-)
The problem with the plugin approach here is that it's monkeypatching
a pretty fundamental method, and undocumented, method. And that makes
it fragile; if the implementation of initialize ever changes (and
there's no reason that it shouldn't), the plugin breaks.
Still, if it's the only game in town...
Ben Munat
Hmm, yeah that's a good point... always the risk with monkeypatching.
So, Piers, you're the one that started this whole thing... if it really is safe to do:
class MyModel < ActiveRecord::Base
def initialize(attrs = {}, &block)
super
# all my initialization stuff here
end
end
and this is documented in the AR::Base docs, then this does kind of seem like enough. I mean, it's
what I did all the time in my java life.
I'd just swear that someone told me the holy beasts of doom would swarm down upon me if I ever tried
to override AR::Base#initialize.
Ben
Pratik
def initialize_with_defaults(attributes = nil, &block)
initialize_without_defaults(attributes) do
self.some_attribute = 'whatever' # Setting default value
yield self if block_given?
end
end
alias_method_chain :initialize, :defaults
Looks like it's working quite well.
Thanks,
Pratik
Piers Cawley
Yeah, I think I've heard the Holy Beasts of Doom argument as well, I
just can't for the life of me remember where. I have certainly been
bitten by trying to set the defaults before calling super though
(because, dammit, that's the logical place to set defaults). Thinking
about it, the way to do that would be something like:
def initialize(attrs = {}, &block)
super attrs.reverse_merge(:default => 'this'), &block
end
but then you make yourself a hostage to fortune that everyone's going
to use symbols as keys. The post super approach to setting defaults
has some edge case problems too in cases where nil is a legal value
for some attribute (yeah, it's a weird edge case, but an edge case all
the same). You end up with ugly code like:
def initialize(attrs = {}, &block)
super
unless attrs.has_key?(:content) || attrs.has_key?('content')
self.content = "Write something here"
end
end
The issue is, I think, that ActiveRecord::Base#initialize is doing two
different 'sorts' of things: class invariant metadata initialization,
and instance specific initialization. You might compose the method as:
def initialize(attributes = nil, &block)
initialize_metadata
initialize_instance(attributes, &block)
end
def initialize_metadata
@attributes = attributes_from_column_definition
@new_record = true
ensure_proper_type
end
def initialize_instance(attributes
self.attributes = attributes unless attributes.nil?
yield self if block_given?
end
Maybe the right thing to do is to implement ActiveRecord::Base.new as:
class ActiveRecord::Base
def self.new(attributes = nil, &block)
returning(self.allocate) do |instance|
instance.initialize_metadata
instance.initialize(attributes, &block)
end
end
def initialize_metadata
@attributes = attributes_from_column_definition
@new_record = true
ensure_proper_type
end
def initialize(attributes)
self.attributes = attributes unless attributes.nil?
yield self if block_given?
end
end
Then anyone who wants to write code that initializes defaults before
the actual attributes are set can do:
def initialize(attributes = nil, &block)
self.content = "Write something here"
super
end
and everybody is happy.
Piers Cawley
>
> I would personally do it as something like :
>
> def initialize_with_defaults(attributes = nil, &block)
> initialize_without_defaults(attributes) do
> self.some_attribute = 'whatever' # Setting default value
> yield self if block_given?
> end
> end
> alias_method_chain :initialize, :defaults
>
> Looks like it's working quite well.
Except in the case where it overrides a value that's already been set
by initialize_without_defaults. I'm not sure that's the behaviour you
want.
Plus, imitating the action of super by hand coding seems like a bit of
a waste of time somehow.
Pratik
> Except in the case where it overrides a value that's already been set
> by initialize_without_defaults. I'm not sure that's the behaviour you
> want.
>
Yeah. initialize_without_defaults is basically AR::Base#initialize -
so yeah, this will override values set by that. If you're already
overriding default values, I'm not sure how this might be a problem.
> Plus, imitating the action of super by hand coding seems like a bit of
> a waste of time somehow.
Well, basically this is supplying a block to AR::Base#initialize which
is executed by "yield self if block_given?" - this protect it against
possible changes that might happen to AR::Base#initialize. It also
allows you to override the default values you've set. So you do
something like :
Model.new do |model|
model.i_dont_want_your_default = 'whatever'
end
It might be useful in case of STI or in controllers where you want to
override default values.
So yeah, it's not really imitating the action of super by hand, but
using proper technique to set defaults I feel. This just feels
cleaner.
Piers Cawley
>
> > Except in the case where it overrides a value that's already been set
> > by initialize_without_defaults. I'm not sure that's the behaviour you
> > want.
> >
>
> Yeah. initialize_without_defaults is basically AR::Base#initialize -
> so yeah, this will override values set by that. If you're already
> overriding default values, I'm not sure how this might be a problem.
Because what you're actually doing having your block override the a
value that's been specified in the attributes hash with your default.
Which probably isn't what you want.
Pratik
def initialize_with_defaults(attributes = nil, &block)
initialize_without_defaults(attributes) do
self.some_attribute ||= 'whatever' # Setting default value if not present
yield self if block_given?
end
end
alias_method_chain :initialize, :defaults
On 7/24/07, Piers Cawley <pdca...@bofh.org.uk> wrote:
Piers Cawley
>
> Then do :
>
> def initialize_with_defaults(attributes = nil, &block)
> initialize_without_defaults(attributes) do
> self.some_attribute ||= 'whatever' # Setting default value if not present
> yield self if block_given?
> end
> end
> alias_method_chain :initialize, :defaults
I think we're starting back around the circle again, but what happens
when the default value is 'true' and the arguments specify 'false'.
And, then, when you've solved that one, what happens when 'nil' is a
legal value, but the default value is non nil?
Pratik
class Item < ActiveRecord::Base
def initialize_with_defaults(attrs = nil, &block)
initialize_without_defaults(attrs) do
setter = lambda { |key, value| self.send("#{key}=", value)
unless !attrs.nil? && attrs.keys.map(&:to_s).include?(key) }
setter.call('scheduler_type', 'hotseat')
yield self if block_given?
end
end
alias_method_chain :initialize, :defaults
end
Can you please point out situations where this will fails ?
Thanks !
-Pratik
On 7/24/07, Piers Cawley <pdca...@bofh.org.uk> wrote:
>
Piers Cawley
>
> Umm..I think that leaves us with something like:
>
> class Item < ActiveRecord::Base
> def initialize_with_defaults(attrs = nil, &block)
> initialize_without_defaults(attrs) do
> setter = lambda { |key, value| self.send("#{key}=", value)
> unless !attrs.nil? && attrs.keys.map(&:to_s).include?(key) }
> setter.call('scheduler_type', 'hotseat')
> yield self if block_given?
> end
> end
> alias_method_chain :initialize, :defaults
> end
>
> Can you please point out situations where this will fails ?
Whenever I you try to understand it?
The 'rewrite' new approach means you have an easily overrideable
'initialize' that works with all the edge cases you're taking account
of. Frankly, I'd much rather be able to write:
def initialize(attrs = nil, &block)
self.body = "Write something here"
super
end
and know it's going to work. Your mileage may vary.
Piers Cawley
Oh yes, I'll bet dollars to doughnuts that the rewrite new approach is
the faster of the two as well.
Pratik
You don't have to lean on sarcasm when you have nothing constructive
to say ( especially after you pointed out deftect in implementation
and I corrected them ). May be you can focus on submitting a patch.
patch -p0 speaks a lot louder than your sad grammar.
AR:Base#initialize is supposed to be an internal method. And it should
be flexible enough to be changed in future for whatever reasons.
ara.t.howard
On Jul 23, 2007, at 7:14 PM, Pratik wrote:
> Why add a new callback when all you need is 4 lines of documentation
> on how to do it ?
because you can say that about all of the callbacks.
a @ http://drawohara.com/
--
we can deny everything, except that we have the possibility of being
better. simply reflect on that.
h.h. the 14th dalai lama
Pratik
before/after_add/remove association collection callbacks ?
Anyways, I'm off this thread unless we see some patches. That makes
arguing a lot easier :-)
ara.t.howard
On Jul 22, 2007, at 4:39 PM, Michael Koziarski wrote:
>
> I can buy your argument, but if I do so, there doesn't seem to be a
> need for after_initialize:
>
> def initialize(attrs)
> do_stuff
> super
> end
>
> All we'd be doing is reinventing an initialize method which hid the
> arguments. Doesn't seem particularly necessary, especially at the
> cost of backwards compatibility for the people who use
> after_initialize at present.
>
the trick, of course, is that attrs are sometimes nil - also we
(client code) don't know what/if any block might be, or not be,
for... i just had this issue last week and here is how i solved it
class Agent < ActiveRecord::Base
def initialize options, &block
super((options || {}).reverse_merge!(defaults), &block)
end
def defaults
Hash.new
end
end
this is used in an STI situation so i also have code like this
class Group < Agent
def defaults
{
'allow_login' => false,
}
end
end
class User < Agent
def defaults
{
'allow_login' => true,
}
end
end
and this seems to be working well. so my own suggestion wouldn't be
so much for another callback but a way to simply set the defaults for
a record, perfering a method over a hash so that current object state
and arbitrary logic might affect the result, Time.now, for instance.
also, i strongly agree that this is a giant hole in the ar lifecycle.
kind regards.
ara.t.howard
On Jul 22, 2007, at 4:39 PM, Michael Koziarski wrote:
>
> I can buy your argument, but if I do so, there doesn't seem to be a
> need for after_initialize:
>
> def initialize(attrs)
> do_stuff
> super
> end
one other nit-pick: it's somewhat archaic and java like to
__require__ client code to call super, one slip of the keyboard and
someone is bug finding...
it's much nicer, imho, to relieve the client code from that burden via
class ActionRecord::Base
def self.new *args, &block
returning(allocate) do |object|
object.instance_eval do
internal_initialize *args, &block # all real work here
initialize *args, &block # default does nothing
end
end
end
end
basically just following the paradigm whereby designing classes that
a made to be inherited with any work in initialize is less that
ideal... food for thought.
cheers.
ara.t.howard
On Jul 24, 2007, at 11:15 AM, Pratik wrote:
>
> Umm...what would you suggest as alternative for
> before/after_add/remove association collection callbacks ?
i'm not suggesting an alternative, i saying that just because docs
*can* take the place of easy code doesn't mean they should. i'd just
hate to see this in the docs:
# this is how to write a 'before save' filter in rails
#
# save = method :save
#
# define_method :save do |*argv|
# p 'before_save'
# save.call *argv
# end
the declarative style of ar is a wonderful thing - better the
leverage it rather than use docs imho.
regards.
Pratik
being false/nil etc. And I assumed the solution to be as simple as
overriding initialize and calling super as the first statement.
On 7/24/07, ara.t.howard <ara.t....@gmail.com> wrote:
>
>
ara.t.howard
On Jul 24, 2007, at 11:15 AM, Pratik wrote:
> Anyways, I'm off this thread unless we see some patches. That makes
> arguing a lot easier :-)
--- vendor/rails/activerecord/lib/active_record/base.rb.org
2007-07-24 11:37:07.000000000 -0600
+++ vendor/rails/activerecord/lib/active_record/base.rb 2007-07-24
11:40:32.000000000 -0600
@@ -1502,10 +1502,14 @@
@attributes = attributes_from_column_definition
@new_record = true
ensure_proper_type
- self.attributes = attributes unless attributes.nil?
+ self.attributes = (attributes ||
{}).stringify_keys.reverse_merge!(defaults.stringify_keys)
yield self if block_given?
end
+ def defaults
+ Hash.new
+ end
+
# A model instance's primary key is always available as model.id
# whether you name it the default 'id' or set it to something
else.
def id
Piers Cawley
Except that, when you're setting defaults, you really would like to
interpose some behaviour between the bit that gets the object to the
point where the various setter/getter methods will work and the bit
where ActiveRecord calls self.attribs = ...
One option would be to override ActiveRecord::Base.allocate to return
an object that's been got to that point, but then you'd be doing a
chunk of work that then gets thrown away whenever an object is made
through AR::Base.instantiate.
Sadly, initialize tends to be one of those warts on an API where you
really can't get away with requiring client code to call super. You
*could* add #before_initialize and #after_initialize hooks, but then
we're back where we started...
ara.t.howard
On Jul 24, 2007, at 12:00 PM, Piers Cawley wrote:
> Except that, when you're setting defaults, you really would like to
> interpose some behaviour between the bit that gets the object to the
> point where the various setter/getter methods will work and the bit
> where ActiveRecord calls self.attribs = ...
yeah i agree completely. my impl just happened to skin my own cat
quite nicely at the time, but a more general solution would do just
that...
before_initialize and after_initialize don't seem to bad after all eh?
kind regards.
Piers Cawley
>
> > Except that, when you're setting defaults, you really would like to
> > interpose some behaviour between the bit that gets the object to the
> > point where the various setter/getter methods will work and the bit
> > where ActiveRecord calls self.attribs = ...
>
> yeah i agree completely. my impl just happened to skin my own cat
> quite nicely at the time, but a more general solution would do just
> that...
The trouble with patching a framework is that you have to account for
every kind of cat under the sun. At the very least, you shouldn't make
it so that some cats are completely unskinnable.
Anyhoo. Here's my patch, complete with a new test. I've probably put
methods in the wrong place, of course, but it does at least work. It's
against the ActiveRecord in Rails 1.2.3, but unless things have
changed quite dramatically it should apply reasonably cleanly to the
edge I think.
Index: test/base_test.rb
===================================================================
--- test/base_test.rb (revision 6690)
+++ test/base_test.rb (working copy)
@@ -47,6 +47,13 @@
attr_accessible :phone_number
end
+class DefaultedPerson < LoosePerson
+ def initialize(*args, &block)
+ self.first_name = 'Bob'
+ super
+ end
+end
+
class Booleantest < ActiveRecord::Base; end
class Task < ActiveRecord::Base
@@ -365,6 +372,11 @@
assert_raises(ActiveRecord::RecordNotFound) { topicReloaded =
Topic.find(99999) }
end
+ def test_initialize_sets_defaults
+ person = DefaultedPerson.new
+ assert_equal("Bob", person.first_name)
+ end
+
def test_initialize_with_attributes
topic = Topic.new({
"title" => "initialized from attributes", "written_on" =>
"2003-12-12 23:23"
Index: lib/active_record/base.rb
===================================================================
--- lib/active_record/base.rb (revision 6690)
+++ lib/active_record/base.rb (working copy)
@@ -453,6 +453,13 @@
end
end
+ def new(attributes = nil, &block) #:nodoc:
+ returning(allocate) do |instance|
+ instance.send(:initialize_metadata)
+ instance.send(:initialize, attributes, &block)
+ end
+ end
+
# Finds the record from the passed +id+, instantly saves it
with the passed +attributes+ (if the validation permits it),
# and returns it. If the save fails under validations, the
unsaved object is still returned.
#
@@ -1502,11 +1509,14 @@
# In both instances, valid attribute keys are determined by the
column names of the associated table --
# hence you can't have attributes that aren't part of the table columns.
def initialize(attributes = nil)
+ self.attributes = attributes unless attributes.nil?
+ yield self if block_given?
+ end
+
+ def initialize_metadata #:nodoc:
@attributes = attributes_from_column_definition
@new_record = true
ensure_proper_type
- self.attributes = attributes unless attributes.nil?
- yield self if block_given?
end
# A model instance's primary key is always available as model.id
Ben Munat
+1
Ben
PS: however, not sure what rev you're diffing against but in head there's another line between the
self.attributes and the yield:
self.class.send(:scope, :create).each { |att,value| self.send("#{att}=", value) } if
self.class.send(:scoped?, :create)
Pratik
ara.t.howard
On Jul 24, 2007, at 1:33 PM, Pratik wrote:
>
> http://pastie.caboo.se/81829
>
> That's my proposed solution.
my take is that is HAS to be a method so that
self.foobar = self.foo + self.bar
which is very hard to do with a hash based mechanism (i realize mine
is hash based too!)
so, to me, it really suggests a callback or post_initialize approach...
2 cts.
Michael Koziarski
> Without cases which
> can't be solved by this pattern, I'm not sure that adding a new
> callback is justified.
Sorry for replying to myself, but the wireless here is a little too
spotty to reply to all the individual threads.
Providing a method to allow static defaults such as:
set_default_values :foo=>'bar'
Is just reimplementing the pre-existing defaults code that we extract
from the columns, so I'm not sure that's the right way to go.
For dynamic defaults, such as 'the default value for the categories on
a blog, is taken from the account's default values' I've not seen a
case that can't be catered for by overriding the initialize method.
Requiring users to call super first is perfectly acceptable to me. To
expect to use methods from a super class without first initializing
that super class seems very strange to me.
Of course, there may well be particular cases where it's genuinely
confusing that something doesn't work, or that the error messages are
misleading, but we can fix those on a case by case basis.
Every feature that we add, becomes something we need to support on an
ongoing basis, for the foreseeable future. So we have to have a
reject by default, or we'll end up with another 'components'.
--
Cheers
Koz
ara.t.howard
i agree nearly 100%. the simplest solution, by far, seem to simply
make after_initialize actually called then. leaving open the door
for before_initialize and remaining backward compatible.
> --
> Cheers
>
> Koz
Piers Cawley
>
> On 7/23/07, Michael Koziarski <mic...@koziarski.com> wrote:
>
> > Without cases which
> > can't be solved by this pattern, I'm not sure that adding a new
> > callback is justified.
>
> Sorry for replying to myself, but the wireless here is a little too
> spotty to reply to all the individual threads.
>
> Providing a method to allow static defaults such as:
>
> set_default_values :foo=>'bar'
>
> Is just reimplementing the pre-existing defaults code that we extract
> from the columns, so I'm not sure that's the right way to go.
>
> For dynamic defaults, such as 'the default value for the categories on
> a blog, is taken from the account's default values' I've not seen a
> case that can't be catered for by overriding the initialize method.
>
> Requiring users to call super first is perfectly acceptable to me. To
> expect to use methods from a super class without first initializing
> that super class seems very strange to me.
class Foo
attr_accessor :bibble
end
class Bar < Foo
def initialize
# Look ma, I'm using a superclass method before I initialize it!
self.bibble = "Pling!"
super
end
end
Happens all the time. The current ActiveRecord behaviour is weird
because it breaks that expectation.
ara.t.howard
On Jul 24, 2007, at 2:50 PM, Michael Koziarski wrote:
>
> For dynamic defaults, such as 'the default value for the categories on
> a blog, is taken from the account's default values' I've not seen a
> case that can't be catered for by overriding the initialize method.
>
> Requiring users to call super first is perfectly acceptable to me. To
> expect to use methods from a super class without first initializing
> that super class seems very strange to me.
>
so i'm hacking on this right now, and the issue with this is as
follows, you basically have to do something like this:
def initialize options = nil, &block
super
if options
options.to_options!
unless options.has_key? :some_column
self.some_column = 'some_default_value'
end
end
end
to be robust because otherwise you will clobber a call like this
Module.new 'some_column' => nil
where the user explicitly asks for a column to be nil, or false for
that matter, or used a string key, etc. you can't just check the
value from the db either since that may be false, etc. my point is
that, to me, 'default' in this context means the following: set it
iff the user did not specify it. clobbering, or setting the value
only because it's not a truth value doesn't fit the bill here and
testing whether it was passed in seems a little verbose for what
should be a very easy task.
maybe i am making it over complicated?
Michael Koziarski
> because it breaks that expectation.
You're describing the current behaviour of active record and it's use
of method_missing to trigger generate_read_methods, and the lack of
generate write methods. It leads to other crazy stuff like
>> Post.public_instance_methods.grep(/body/)
=> []
>> Post.find(1).body
=> "Howdy World"
>> Post.public_instance_methods.grep(/body/)
=> ["body?", "body"]
I agree that this is counter intuitive, if you have a fix for this,
I'd definitely be interested in applying it, the current behaviour
causes problems here and in other cases. However this is the problem
that should be fixed, not adding a callback to work around existing
confusing implementation of our attribute related methods.
--
Cheers
Koz
ara.t.howard
On Jul 24, 2007, at 2:50 PM, Michael Koziarski wrote:
>
> Sorry for replying to myself, but the wireless here is a little too
> spotty to reply to all the individual threads.
>
> Providing a method to allow static defaults such as:
>
> set_default_values :foo=>'bar'
>
> Is just reimplementing the pre-existing defaults code that we extract
> from the columns, so I'm not sure that's the right way to go.
>
> For dynamic defaults, such as 'the default value for the categories on
> a blog, is taken from the account's default values' I've not seen a
> case that can't be catered for by overriding the initialize method.
>
> Requiring users to call super first is perfectly acceptable to me. To
> expect to use methods from a super class without first initializing
> that super class seems very strange to me.
>
> Of course, there may well be particular cases where it's genuinely
> confusing that something doesn't work, or that the error messages are
> misleading, but we can fix those on a case by case basis.
>
> Every feature that we add, becomes something we need to support on an
> ongoing basis, for the foreseeable future. So we have to have a
> reject by default, or we'll end up with another 'components'.
okay, this isn't a patch but it's what i'm using internally and a
suggestion for an impl:
class ActiveRecord::Base
alias_method '__initialize__', 'initialize'
def initialize options = nil, &block
returning( __initialize__(options, &block) ) do
options ||= {}
options.to_options!
defaults = self.class.defaults || self.defaults || Hash.new
(defaults.keys - options.keys).each do |key|
value = defaults[key]
case value
when Proc
value = instance_eval &value
when Symbol
value = send value
end
send "#{ key }=", value
end
end
end
def self.defaults *argv
@defaults = argv.shift.to_hash if argv.first
return @defaults if defined? @defaults
end
def defaults *argv
@defaults = argv.shift.to_hash if argv.first
return @defaults if defined? @defaults
end
end
while this might seem overly complicated at first it makes all this
behave correctly
class C < ActiveRecord::Base
defaults :foo => 42
end
p C.new.foo #=> 42
p C.new('foo' => 42.0) #=> 42.0
class C < ActiveRecord::Base
defaults :foo => :bar
def bar() 42 end
end
p C.new.foo #=> 42
class C < ActiveRecord::Base
defaults :foo => lambda{ 42 }
end
p C.new.foo #=> 42
class C < ActiveRecord::Base
defaults :foo => lambda{ bar }
def bar() 42 end
end
p C.new.foo #=> 42
also note that the defaults are set *after* the 'normal'
initialization had been executed - allowed things like
class C
default :foobar => lambda{ foo + bar }
end
kind regards.
Piers Cawley
Um... the patch I posted doesn't actually add a callback, it just
rejigs things so that
def initialize(attrs = nil)
self.some_attrib = 'a default value'
super
end
works as well as
def initialize(attrs = nil)
super
self.some_attrib ||= 'a default value'
end
As for the public_instance_methods issue, when does a class find out
about its columns? Assuming columns is known about early enough before
any records are fetched from the database, it should be possible to
fake out public_instance_methods.
It's just a matter of finding someone who has an appropriately shaped tuit.
Michael Koziarski
> rejigs things so that
>
> def initialize(attrs = nil)
> self.some_attrib = 'a default value'
> super
> end
>
> works as well as
>
> def initialize(attrs = nil)
> super
> self.some_attrib ||= 'a default value'
> end
I'm not sold on the need for this in and of itself but if someone else
feels it's justified I wouldn't object to them applying it. I'd
prefer to see work done towards the correct fix than put something in
that wouldn't be needed in the long term, and if we have currently
motivated people, all the better :).
> As for the public_instance_methods issue, when does a class find out
> about its columns? Assuming columns is known about early enough before
> any records are fetched from the database, it should be possible to
> fake out public_instance_methods.
The columns are lazily retrieved, but once they're retrieved it should
be possible to generate all the accessors and mutators. You don't
want to retrieve them too early as the database isn't always
available, however it's definitely something that a motivated
experimenter could investigate in a relatively short period of time.
--
Cheers
Koz
ara.t.howard
On Jul 24, 2007, at 2:50 PM, Michael Koziarski wrote:
> Every feature that we add, becomes something we need to support on an
> ongoing basis, for the foreseeable future. So we have to have a
> reject by default, or we'll end up with another 'components'.
i hear that. i put up a summary of what i'm doing now in my code,
for posterity, at:
http://drawohara.tumblr.com/post/6677354
maybe it will inspire people or at least serve as an example of what
people hate ;-)
Pratik
Any inputs for that ?
Thanks.
Piers Cawley
>
> > Um... the patch I posted doesn't actually add a callback, it just
> > rejigs things so that
> >
> > def initialize(attrs = nil)
> > self.some_attrib = 'a default value'
> > super
> > end
> >
> > works as well as
> >
> > def initialize(attrs = nil)
> > super
> > self.some_attrib ||= 'a default value'
> > end
>
> I'm not sold on the need for this in and of itself but if someone else
> feels it's justified I wouldn't object to them applying it. I'd
> prefer to see work done towards the correct fix than put something in
> that wouldn't be needed in the long term, and if we have currently
> motivated people, all the better :).
The thing is, I actually think that the fix that makes initialize
overloadable in the same way as pretty much every other ruby class out
there _is_ the correct thing, regardless of whether there there are
any class methods in place to set defaults in a more declarative
fashion. Given a straightforwardly overloadable initialize + a
declarative setup, the user has the choice of how they implement
defaults, either by overloading or by using 'default :foo => whatever'
type declarations.
Chris Cruft
but not for initialize. I've been using something like Piers'
approach, which has always worked for me:
def initialize(attrs = {}, &block)
super attrs.reverse_merge(default_attributes_hash), &block
more_initialization_dependent_on_instance_already_being_built
end
I see the problem with string-versus-symbol keys, but Rails has had
that problem for a long time as well as tolerable solutions
(HashWithIndifferentAccess, or the manual equivalent).
Unless somebody has some good evidence of why overriding initialize is
inherently dangerous (and not just tricky w.r.t. before and after
super), I would vote for leaving the callbacks as they are.
On Jul 24, 6:40 am, "Piers Cawley" <pdcaw...@bofh.org.uk> wrote:
> On 24/07/07, Ben Munat <bmu...@gmail.com> wrote:
>
> > So, Piers, you're the one that started this whole thing... if it really is safe to do:
>
> > class MyModel < ActiveRecord::Base
> > def initialize(attrs = {}, &block)
> > super
> > # all my initialization stuff here
> > end
> > end
>
> > and this is documented in the AR::Base docs, then this does kind of seem like enough. I mean, it's
> > what I did all the time in my java life.
>
> > I'd just swear that someone told me the holy beasts of doom would swarm down upon me if I ever tried
> > to override AR::Base#initialize.
>
> Yeah, I think I've heard the Holy Beasts of Doom argument as well, I
> just can't for the life of me remember where. I have certainly been
> bitten by trying to set the defaults before calling super though
> (because, dammit, that's the logical place to set defaults). Thinking
> about it, the way to do that would be something like:
>
> def initialize(attrs = {}, &block)
> super attrs.reverse_merge(:default => 'this'), &block
> end
>
> but then you make yourself a hostage to fortune that everyone's going
> to use symbols as keys. The post super approach to setting defaults
> has some edge case problems too in cases where nil is a legal value
> for some attribute (yeah, it's a weird edge case, but an edge case all
> the same). You end up with ugly code like:
>
> def initialize(attrs = {}, &block)
> super
> unless attrs.has_key?(:content) || attrs.has_key?('content')
> self.content = "Write something here"
> end
> end
>
> The issue is, I think, that ActiveRecord::Base#initialize is doing two
> different 'sorts' of things: class invariant metadata initialization,
> and instance specific initialization. You might compose the method as:
>
> def initialize(attributes = nil, &block)
> initialize_metadata
> initialize_instance(attributes, &block)
> end
>
> def initialize_metadata
> @attributes = attributes_from_column_definition
> @new_record = true
> ensure_proper_type
> end
>
> def initialize_instance(attributes
> self.attributes = attributes unless attributes.nil?
> yield self if block_given?
> end
>
> Maybe the right thing to do is to implement ActiveRecord::Base.new as:
>
> class ActiveRecord::Base
> def self.new(attributes = nil, &block)
> returning(self.allocate) do |instance|
> instance.initialize_metadata
> instance.initialize(attributes, &block)
> end
> end
>
> def initialize_metadata
> @attributes = attributes_from_column_definition
> @new_record = true
> ensure_proper_type
> end
>
> def initialize(attributes)
> self.attributes = attributes unless attributes.nil?
> yield self if block_given?
> end
> end
>
> Then anyone who wants to write code that initializes defaults before
> the actual attributes are set can do:
>
> def initialize(attributes = nil, &block)
> self.content = "Write something here"
> super
> end
>
> and everybody is happy.
Hendy Irawan
Largely the case is simply setting defaults with "if new_record?". Since it's 80% of the job that can be handled with (ficitional) 20% code such as:
def after_new_record
country = 'United States'
is_administrator? = false
end
I think that's nice :) (since it doesn't use a dynamic trigger, it should pose not very big overhead?)
-- Hendy Irawan www.hendyirawan.com