link_to / form_for doesn't work for singular resource

[Godisemo]

In my routes I have

resource :company

and in my views

<% form_for(@store) do |f| %>
    …
<% end %>

When

new action gets an error when rendering

undefined method hash_for_companies_path' for #Module:0x00000102bfd3f0`

The stack trace reviels that the problem occurs in

lib/action_dispatch/routing/polymorphic_routes.rb:133:in polymorphic_url'`

edit action can render without errors but the forms url is /company.4 where 4 is the id of the company.

This seems to be a bug that has been around for a very long time so I think it's time to fix it.

[pixeltrix]

You're right the error has been around for a long while however a clean solution doesn't readily represent itself. The polymorphic_url helper has no 'intelligence' in how it operates - it has no information about what you've declared as resources in your routes.rb. All it has to go on it the name of the model and how that maps to the named url helpers.

In your case when passing a new Company instance it tries to map to the plural path as that's the standard route for the create action. The question is how to determine a singular resource from a regular resource - suggestions are welcome. The best solution I can think of is to try a respond_to? first and if it doesn't then try the singular option - I'll take a look at it.

[gucki]

Ah, didn't know there's already a ticket for htis. So here's my post from the mailing list:

What about making the polymorphic routing in rails to try a plural route first. If it fails, try a singleton route. If it fails, raise.

This is not the nicest and most performant way (the result could be cached in production so it doesn't really matter in the end) but I guess it should solve most cases. Or is this too much magic?

[wildchild]

I am wondering why this ticket pointed to 3.2 milestone?

[pixeltrix]

Because it’s not a blocker for the 3.1 release and it may require too many changes to go into a patch release.

[mikeni]

i think this bug should be written in rails guides somewhere.

for anyone using a singular resource, they will eventually find themselves on this bug page. better to announce first

[maximeg]

Still occurring.

[mikeni]

for now i just use scope =(

[phantomwhale]

Yep, my first time with the singular resource form, and I walked smack into this bug. Would be great to see a fix in an upcoming version - don't see any target milestone, so I presume it's not under active development at the moment ?

[kjakub]

Yep, here too, my first time with singular resource, lost 3h coz i was blaming myself like a first

[glebm]

+1

[felipero]

I believe respond_with has the same problem.

[pixeltrix]

@felipero yes it will - as will anything where you can pass a model instance and get a url.

I haven't forgotten about this issue, it's just trickier than it seems to fix. The problem is there is no easy way to discern whether a model maps to a singular or a regular resource url. Checking for the presence of a collection url doesn't work as the resource may have been specified with :except => :index and trying to rescue route generation errors doesn't work because passing an instance to a singular resource url helper will generate a url with the format set to the id and no exception.

The only apparent way I can think of doing it is by creating a data structure that keeps a track of whether a set of routes is a singular resource and then querying that data structure in polymorphic_url. However before I or someone else can go ahead with that there needs to be a discussion whether we want the overhead of that structure.

[felipero]

Well, it is a non-blocking issue. Thanks for you prompt answer. :)

[zelig]

I think someone ought to patch this. My thinking would not be along the lines of trying routes to fail (and fall back on the plural as in the comments in the rails issue) but go after association reflections.

Can we make the assumption that behind every nested resource form there is an activerecord association? No but we can test it

for form_for [a, b] (of classes A, B) if there is a model class A with has_many assoc to B, we can safely assume, expect that routes are set up to support a plural ("a_bs_path") for create. If there is a has_one we can safely assume, expect that routes are set up to support a singular ("a_bs_path"). If we find none, we fall back to the plural as the status quo educated guess. My 2 cents. Is this against modularity and decoupling MVC?

[pixeltrix]

@zelig firstly, I think anything like that is tying your models to your routes too much. Secondly, what about non-AR models, and finally it doesn't fix the problem for top-level singular resources, e.g;

class Basket < ActiveRecord::Base
end

resource :basket

There's nothing in the model that's going to help with that - the only way it to store the fact that it's a singular resource and query that when generating the url. Adding metadata to the model won't help as a model may sometimes be a singular resource (/account) or a multiple resource (/admin/users/1).

[zelig]

Surely this doesn't solve all problems but in the special case of nested AR resources, plurality can be reasonably guessed using model class associations, for others (i.e., top-level resource or non-AR models or models with implicit associations) the plural is a decent guess. This sounds to me a simple improvement but maybe the caveat is that it is not the right solution.

So where would you store resource singularity so that polyorphic_url could access this info? In case people are interested in fixing this ugly duckling.

[jakevose]

Add me to the list of people who just lost a bunch of time trying to figure out the "right" way to work around this and thinking myself crazy before stumbling upon this obscure reference to a long-standing, known bug.

[pjg]

Add me to the list of people who just lost a bunch of time trying to figure out the "right" way to work around this and thinking myself crazy before stumbling upon this obscure reference to a long-standing, known bug.

And me as well.

[pjg]

To anyone stumbling upon this issue, the workaround is quite simple. Given:

resource :billing_info

You generate the form_for call like this:

= form_for @billing_info, url: billing_info_path do |f|

[gmile]

@pjg one of the nice features of form_for is to generate a proper url, based on object's persistance state, i.e. autodetect a path with either _create or _update affixes. That said, i'm unsure if the solution you propose will handle both cases at the same time. Will it?

[pjg]

Yeah, it does handle both :create and :update actions.

[asymmetric]

It would be really nice to have this fixed, or at least to have the problem prominently documented somehow. Just spent 2hrs thinking it was an error on my end!

[jlbenc]

Wasted 2+ hours hunting this issue down, :(

Thanks @pjg for your workaround! 👍

[gaurish]

Today, I faced issue too. solved via workaround given Thanks @pjg 👍

[hasghari]

I agree with @mikeni that this bug should be documented in the Rails guides.

[nazar]

Thanks @pijg for saving me an hour or three

[gustavokloh]

+1 for this issue!

And one more thanks @pjg

[philipgiuliani]

Why a bug like this isn't fixed yet?

[notrab]

@philipgiuliani I'm sure you'll discover, judging by the comments above, there isn't any easy way to do this without polluting other classes within Rails for this behaviour.

The workaround is quite simple and to be honest, I don't see a problem. Throw it in the docs and just be done with it. Adding url: to the form declaration isn't so hard work for singular routes.

😴

[rails-bot]

This issue has been automatically marked as stale because it has not been commented on for at least
three months.

The resources of the Rails team are limited, and so we are asking for your help.

If you can still reproduce this error on the 4-2-stable, 4-1-stable branches or on master,
please reply with all of the information you have about it in order to keep the issue open.

Thank you for all your contributions.

[rafaelfranca]

See #17066 (comment)

[schuetzm]

@rafaelfranca Care to explain why you closed this issue? The comment you referenced says it is still open "because it's a difficult problem to solve". Has it finally been fixed?

[rafaelfranca]

Oops! Looks like I linked to the wrong comment. I could not find the comment that I explained that it is fine to use the :url option and that this problem is not something that we need to fix.

#23138 may fix this issue introducing a new API.

[matkoniecz]

Why it is closed? #23138 is not merged and it is explicitly documented as existing bug in http://guides.rubyonrails.org/v4.1/routing.html

[azul]

@matkoniecz - it's documented as you said including a workaround with the url param. And there is a bug that may fix this in the longer run. I think that is why it's closed. #23138 will track the progress on the fix now.

[IamNaN]

For anyone landing here via google, this is what got implemented: http://edgeguides.rubyonrails.org/routing.html#singular-resources

In short, add something like resolve('User') { [:user] } after you define the singular resource in routes.rb, where 'User' is the class name and :user is the resource name.

[tomrossi7]

Couldn't this be fixed in the latest rails implementation by changing the behavior of form_with?