# A has\_many :bs; what is the Rails idiom to save @a and @.bs in one transaction?

**URL:** <https://www.ruby-forum.com/t/a-has-many-bs-what-is-the-rails-idiom-to-save-a-and-bs-in-one-transaction/151629>\
**Category:** Rails\
**Created:** [October 23, 2008, 4:02pm UTC](https://www.ruby-forum.com/t/a-has-many-bs-what-is-the-rails-idiom-to-save-a-and-bs-in-one-transaction/151629 "2008-10-23T16:02:27Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![dewie](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/ac8455/32.png) [@dewie](https://www.ruby-forum.com/u/dewie)\
**Post date:** [October 23, 2008, 4:02pm UTC](https://www.ruby-forum.com/t/a-has-many-bs-what-is-the-rails-idiom-to-save-a-and-bs-in-one-transaction/151629/1 "2008-10-23T16:02:27Z")

</div>

Suppose I have two models:

class A \< ActiveRecord::Base  
has\_many :bs  
end

class B \< ActiveRecord::Base  
belongs\_to :a  
end

Now, if I say in a controller  
b1 = @a.build(params[:b])  
I can build a new B associated to an A.

If the @ is also new, I need to also do this  
b1.a = @a  
but if the @a is not new, then its id is filled in for me.

(1) Why? Even though there is no @a.id to set the b1.a to, there is  
an @a object that I can set b1.a to, and I do. Couldn’t build do this  
for me so @a.build(…) would work in both cases?

Either way, if I now say  
@a.save!  
Then not only are the values of @a that may have been updated saved,  
the new object b1 is also saved to the database. Not only that, this  
all happens in the same transaction. Yea!

However, instead if I were to update an existing b as follows:  
b1 = @a.bs[1].update(params[:b])  
and then  
@a.save!  
then this time, the updated values of b1 are not saved to the  
database.

(2) Why? Wouldn’t it be cleaner to update the associated b values  
into the database as well during the save of the @a ?

Further, if I want the @a and the associated @a.bs to all be sent to  
the database in the same transaction, as far as I can tell, I have the  
following choices.

In my controller save the @a and @a.bs in one transaction:  
ActiveRecord::Base.transaction do  
@a.bs.each do |b|  
b.save!  
end  
@a.save!  
end

Or perhaps to be more elegant I could push all that into a method  
save\_it! on class A.

Or I can try to figure out how to get the @a.bs to save in the same  
transaction as the @a. I suppose I can do that as follows:

class A \< ActiveRecord::Base  
before\_save do |a|  
bs.each do |b|  
b.save!  
end  
end  
end

This is a bit better, but – amazingly – I have somehow at times  
gotten into an infinite loop doing this, though it was a bit more  
complex than that. Even though the documentation says that if I save  
on the belongs\_to side, the object belonged-to will not be saved, I  
had a case where saving the associated objects like this ended up  
recursing on itself. Is this idiom supposed to work and was I  
therefore getting an infinite loop because (a) I was not doing it  
quite right, or (b) is this idiom not good.

(3) What is the right idiom for this situation, where I want to save  
the @a and the associated bs in one transaction?

Daniel