722653 - Pipes (|) in the URL bar are decoded for display and not re-encoded when submitting the URL again

Status: RESOLVED INVALID

(Firefox :: Address Bar, defect)

Description

[dg]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0a1) Gecko/20120130 Firefox/12.0a1
Build ID: 20120130031142

Steps to reproduce:

Install FF 12.0a1 (2012-01-30)
Open https://www.google.com/?q=a%7Cb
Note that the URL bar now shows https://www.google.com/?q=a|b
Click into the URL bar, press enter


Actual results:

GET request to https://www.google.com/?q=a|b (you can watch in HttpFox)
Firefox does not encode the | ("pipe") character


Expected results:

GET request to https://www.google.com/?q=a%7Cb
Firefox should encode the | ("pipe") character, according to http://en.wikipedia.org/wiki/Percent-encoding

Comment 1

[[:Aleksej]]

Thanks for filing the bug, sorry it stayed untouched.

Can you reproduce it with a recent version (also it might be helpful if Firefox's internal development tools can be used for that)?

Comment 2

[Dão Gottwald [:dao]]

(In reply to dg from comment #0)
> User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0a1) Gecko/20120130
> Firefox/12.0a1
> Build ID: 20120130031142
> 
> Steps to reproduce:
> 
> Install FF 12.0a1 (2012-01-30)
> Open https://www.google.com/?q=a%7Cb
> Note that the URL bar now shows https://www.google.com/?q=a|b
> Click into the URL bar, press enter
> 
> 
> Actual results:
> 
> GET request to https://www.google.com/?q=a|b (you can watch in HttpFox)
> Firefox does not encode the | ("pipe") character
> 
> 
> Expected results:
> 
> GET request to https://www.google.com/?q=a%7Cb
> Firefox should encode the | ("pipe") character, according to
> http://en.wikipedia.org/wiki/Percent-encoding

That page doesn't actually say that | should be encoded. Instead, it says: "Characters from the unreserved set never need to be percent-encoded. [...] For maximum interoperability, URI producers are discouraged from percent-encoding unreserved characters."

Comment 3

[dg]

(In reply to Dão Gottwald [:dao] from comment #2)
> That page doesn't actually say that | should be encoded. Instead, it says:
> "Characters from the unreserved set never need to be percent-encoded. [...]
> For maximum interoperability, URI producers are discouraged from
> percent-encoding unreserved characters."

I disagree. The pipe symbol is neither in the set of reserved nor the set of unreserved characters (the article gives a complete list of both character sets), thus your quote does not apply. 

RFC2396 even has a section (2.4.3) mentioning the pipe character ("unwise") and continues:
"Data corresponding to excluded characters must be escaped in order to be properly represented within a URI."

The issue is still present in Firefox 23.0 (Ubuntu 13.04) and can be observed as described earlier.