# Question updating linked records **URL:** \ **Category:** Rails\ **Created:** [November 8, 2006, 3:14pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256 "2006-11-08T15:14:51Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 3:14pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/1 "2006-11-08T15:14:51Z")
If you update the id of a linked table in the model, wouldn’t you expect the rails framwork to immediately take this into account? what I mean is: rates table id currency\_id name 1 1 “rate 1” 2 1 “rate 2” currencies table id name 1 “GBP” 2 “USD” 3 “EUR” assume models have the correct belongs\_to and has\_many statements. rate = Rate.find(1) puts rate.currency.name # GBP rate.currency\_id = 2 puts rate.currency.name # still GBP But, rate = Rate.find(2) rate.currency\_id = 2 puts rate.currency.name # USD I understand why this - in the first example, we’ve already retrieved the associated currency record, so it doesn’t re-retrieve it for the second puts. But isn’t this a bug? Shouldn’t rails invalidate the cached associated record when the currency\_id field is changed? Or is this a config setting somewhere? sr ---
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 3:33pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/2 "2006-11-08T15:33:18Z")
i dont think its a bug. did you try this: rate = Rate.find(1) puts rate.currency.name rate.currency\_id = 2 rate.save puts rate.currency.name That _should_ work. And it makes sense that it works like this. Otherwise you could _forget_ to actually update the record in the table, and get strange bugs. Rails forces you to save the changed record so the tables stay in sync with your changes … That’s how i see it. ---
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 4:15pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/3 "2006-11-08T16:15:33Z")
Thorsten L wrote: > i dont think its a bug. > > did you try this: > > rate = Rate.find(1) > puts rate.currency.name > rate.currency\_id = 2 > rate.save > puts rate.currency.name > > That _should_ work. And it makes sense that it works like this. > Otherwise you could _forget_ to actually update the record in the > table, and get strange bugs. Rails forces you to save the changed > record so the tables stay in sync with your changes … > > That’s how i see it. Yes, that will work, but isn’t applicable in an application where you want to allow the user to play around with various options before saving it to the database. If you save after he’s made each change, you’ve got to work out how to get back to the original position should he abandon the update. sr ---
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 4:43pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/4 "2006-11-08T16:43:38Z")
its no bug. in his second example, he did not load the associated records before changing the primary key, so they weren’t cached, but retrieved from the database. However, i think my argumentation from the previous post is not really solid, but i think i just just found the solution. It’s all in the API docs. See rails API FOR: ActiveRecord::Associations::ClassMethods Quote: "Adds the following methods for retrieval and query for a single associated object that this object holds an id to. association is replaced with the symbol passed as the first argument, so belongs\_to :author would add among others author.nil?. " - association(force\_reload = false) - returns the associated object. Nil is returned if none is found. EndofQuote so something like this should work, not tested though rate = Rate.find(1) puts rate.currency.name rate.currency\_id = 2 rate.currency(true) puts rate.currency.name ---
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 5:54pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/5 "2006-11-08T17:54:59Z")
> "Adds the following methods for retrieval and query for a single > associated object that this object holds an id to. association is > replaced with the symbol passed as the first argument, so belongs\_to > :author would add among others author.nil?. " > > - association(force\_reload = false) - returns the associated object. > Nil is returned if none is found. > > EndofQuote > so something like this should work, not tested though > > rate = Rate.find(1) > puts rate.currency.name > rate.currency\_id = 2 > rate.currency(true) > puts rate.currency.name Thorsten, This appears to be just what we’re looking for. We’ll try it out and let you know. sr ---
**Author:** ![Stephen\_Richards](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/b3f665/32.png) [@Stephen\_Richards](https://www.ruby-forum.com/u/Stephen_Richards)\ **Post date:** [November 8, 2006, 4:18pm UTC](https://www.ruby-forum.com/t/question-updating-linked-records/75256/6 "2006-11-08T16:18:01Z")
Thorsten L wrote: > i dont think its a bug. > > did you try this: > > rate = Rate.find(1) > puts rate.currency.name > rate.currency\_id = 2 > rate.save > puts rate.currency.name > > That _should_ work. And it makes sense that it works like this. > Otherwise you could _forget_ to actually update the record in the > table, and get strange bugs. Rails forces you to save the changed > record so the tables stay in sync with your changes … > > That’s how i see it. If that was the case then why would his second example work. Sounds like a (small) bug to me!