Forum: Ruby on Rails path vs. url

When using named RESTful routes, when should one use the
generated ..._path helpers, and when the ..._url helpers?  Say I did
this in routes.rb:

  map.resources :users

Where should I use users_path, new_user_path, etc., and where should I
use users_url, new_user_url...?

Thanks!

Steve

stevemolitor@gmail.com wrote:
> When using named RESTful routes, when should one use the
> generated ..._path helpers, and when the ..._url helpers?  Say I did
> this in routes.rb:
> 
>   map.resources :users
> 
> Where should I use users_path, new_user_path, etc., and where should I
> use users_url, new_user_url...?
> 
> Thanks!
> 
> Steve

The users_url helper generates a URL that includes the protocol and host 
name.  The users_path helper generates only the path portion.

users_url: http://localhost/users
users_path: /users

Mostly you should use the _path flavor. If you need to spec the host or 
protocol (like for talking to another app or service), then use the _url 
flavor.

--
Josh Susser
http://blog.hasmanythrough.com

> Mostly you should use the _path flavor. If you need to spec the host or
> protocol (like for talking to another app or service), then use the _url
> flavor.

Usually the rule is _path in views, _url in controller (where you
mostly use it together with redirect_to).

DHH wrote:
>> Mostly you should use the _path flavor. If you need to spec the host or
>> protocol (like for talking to another app or service), then use the _url
>> flavor.
> 
> Usually the rule is _path in views, _url in controller (where you
> mostly use it together with redirect_to).

and what about to use everytime url?
(i think that the DHH's reply will be: "you're insaaaane"... :) )

David [or anyone], would you mind explaining why you wouldn't want to 
use
the _path form in a controller. I've never done it but I don't 
understand
why it's bad form. Thanks.

RSL

Hi --

On 3/14/07, Josh Susser <rails-mailing-list@andreas-s.net> wrote:
> >
> Mostly you should use the _path flavor. If you need to spec the host or
> protocol (like for talking to another app or service), then use the _url
> flavor.

I guess a case could be made, though, that for REST compliance, you'd
want to use _url throughout, as _path doesn't give you a unique
identifier that can be used to locate a resource outside of this
particular browser session.


David

--
Q. What is THE Ruby book for Rails developers?
A. RUBY FOR RAILS by David A. Black (http://www.manning.com/black)
   (See what readers are saying!  http://www.rubypal.com/r4rrevs.pdf)
Q. Where can I get Ruby/Rails on-site training, consulting, coaching?
A. Ruby Power and Light, LLC (http://www.rubypal.com)

> I guess a case could be made, though, that for REST compliance, you'd
> want to use _url throughout, as _path doesn't give you a unique
> identifier that can be used to locate a resource outside of this
> particular browser session.

I'm not sure I follow? *_path are for views because ahrefs are
implicitly linked to the current URL. So it'd be a waste of bytes to
repeat it over and over.

In the controller, though, *_url is needed for redirect_to because the
HTTP specification mandates that the Location: header in 3xx redirects
is a complete URL.

Hi --

On 3/15/07, DHH <david.heinemeier@gmail.com> wrote:
>
> > I guess a case could be made, though, that for REST compliance, you'd
> > want to use _url throughout, as _path doesn't give you a unique
> > identifier that can be used to locate a resource outside of this
> > particular browser session.
>
> I'm not sure I follow? *_path are for views because ahrefs are
> implicitly linked to the current URL. So it'd be a waste of bytes to
> repeat it over and over.

I'm overstating it (didn't mean to dismiss all relative URLs :-) --
just thinking out loud about a case where the hyperlinks in a response
might be needed to identify resources but the response not contain the
necessary base URL identification.  I'm not coming up with any great
sample scenarios, though.


David

--
Q. What is THE Ruby book for Rails developers?
A. RUBY FOR RAILS by David A. Black (http://www.manning.com/black)
   (See what readers are saying!  http://www.rubypal.com/r4rrevs.pdf)
Q. Where can I get Ruby/Rails on-site training, consulting, coaching?
A. Ruby Power and Light, LLC (http://www.rubypal.com)

Thanks, David. That makes complete sense. I got why you'd want to use 
_path
in the views. I just didn't know that bit about the HTTP specs.

RSL

DHH wrote:
>> Mostly you should use the _path flavor. If you need to spec the host or
>> protocol (like for talking to another app or service), then use the _url
>> flavor.
> 
> Usually the rule is _path in views, _url in controller (where you
> mostly use it together with redirect_to).

I'm getting an error when trying to use a resource_url helper inside my 
controller to render a view.

Here is the problem explained:
http://stackoverflow.com/questions/1150339/rails-2-3-2-trying-to-render-erb-instead-of-haml

DHH wrote:
>> I guess a case could be made, though, that for REST compliance, you'd
>> want to use _url throughout, as _path doesn't give you a unique
>> identifier that can be used to locate a resource outside of this
>> particular browser session.
> 
> I'm not sure I follow? *_path are for views because ahrefs are
> implicitly linked to the current URL. So it'd be a waste of bytes to
> repeat it over and over.
This thread is old but still shows up in top rank in search engines.

Here is my advise: use _url all the time. Yeah it wastes a few bytes, 
but the day you'll add https support to your site it will save you tons 
of headaches. Same applies to images that will end up in RSS feeds. 
Using _path would give them a feed:// protocol.

So save yourself troubles and use _url all the time, period.

Regards,


--
http://www.digiprof.tv - VOD et visioconference

> Here is my advise: use _url all the time. Yeah it wastes a few bytes,
> but the day you'll add https support to your site it will save you tons
> of headaches.

How would it save headaches in case of https?.

Lets say you have apache in front of your rails server and only apache 
is configured for https. And http is disabled in apache.

Now with _url option, the web page gets generated as
<form ... action="http://host/resource/1" ...

And apache passes it along as it as, unless one enables URL rewriting in 
apache to replace http with https.

Where as if it is just _path it gets generated as

<form ...action="/resource/1" ...

This form submission works irrespective of how(http or https) apache is 
configured.

So the earlier recommendation ( i.e., use _path in views _url in 
specific use cases) is still a better one compared to use _url in all 
cases. (no period ;-) )

Here's a reason where I wish I was using _url everywhere.

I've switched my site to almost all ssl except for a google map page 
which needs to be in non-ssl ($10,000 for the ssl version).  On that 
page (just like every other page) are about 15-20 navigation links that 
should point to https/ssl pages even though the map page itself is http.

I wish I wouldn't have to go through and change all those links to point 
to https using _url.  I could instead use ssl_requirement and specify 
all pages should be ssl except for my map page, but I think I'd rather 
make all 15-20 links of navigation use the _url instead and force https 
to avoid an unneeded redirect/latency.  I actually have redirects in 
apache conf file instead of using ssl_requirement, but same idea...why 
redirect when you can specify _url and have less latency?

If you have a solution for me that would relieve my need to have _url 
for all my navigation due to the one http map page, please, let me know.

Kevin


Madhav V. wrote in post #969656:

>
> How would it save headaches in case of https?.
>