News for hackers. Stuff that matters.
Re:URLs Are The Uniform Way to Locate Resources by m0th87 on Wednesday March 31, 2010 @12:36AM (#1230930)
All of your examples still portray uniformity:

> git://github.com/thoughtbot/paperclip.git

You are accessing a resource that is partially identified by the username.

> smtp://user:pass@hostname/domain

The username and password is used for authentication to access the resource, but is part of its unique identification.

> http://news.ycombinator.com/user?id=krainboltgreene

The mistake actually lies with HN, because part of what should be used to identify the resource is instead a param.

> chrome://history/

This still fits because you are accessing a uniquely identified resource. Just because it is user-dependent does not mean it breaks REST.

Re:URLs Are The Uniform Way to Locate Resources by blasdel on Wednesday March 31, 2010 @03:56AM (#1231172)
URL Parameters ≠ RPC

The resource is the whole string between the domain and the fragment identifier. '/' is no different from '?' or '&' — it's all the same once you drop your false-REST preconceptions.

The HTML spec does define a standard for concatenating the name-value pairs from form elements onto target attributes for use in GET requests, but that doesn't reify it into the definition of HTTP, much less the definition of REST.

Re:URLs Are The Uniform Way to Locate Resources by m0th87 on Wednesday March 31, 2010 @10:17AM (#1231816)
Except the use of parameters for resource identification both breaks REST and isn't SEO-friendly. Forms somewhat break this for GET submissions, but do at least adhere for POST.