On Fri, Jul 5, 2013 at 8:21 PM, <e...@racket-lang.org> wrote:
>
>
> 8d55b89 Eli Barzilay <e...@racket-lang.org> 2013-07-05 20:20
> :
> | Further explain silent acceptance of unencoded characters.
> |
> | Also highlight the two tests that use that.
> :
> M pkgs/racket-pkgs/racket-test/tests/net/url.rkt | 7 +++++--
> M racket/lib/collects/net/uri-codec.rkt | 9 ++++++---
>
> =====[ Overall Diff ]===================================================
>
> pkgs/racket-pkgs/racket-test/tests/net/url.rkt
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> --- OLD/pkgs/racket-pkgs/racket-test/tests/net/url.rkt
> +++ NEW/pkgs/racket-pkgs/racket-test/tests/net/url.rkt
> @@ -44,7 +44,7 @@
> => (->vec expected)))
>
> (provide tests)
> -(module+ main (tests))
> +(module+ main (test do (tests)))
> (define (tests)
> (test
> ;; Test the current-proxy-servers parameter can be set
> @@ -254,9 +254,12 @@
> (test-s->u #("mailto" #f #f #f #f (#("ro...@racket-lang.org")) () #f)
> "mailto:ro...@racket-lang.org";)
>
> + ;; The following two tests are not really correct: they rely on the URL
> + ;; decoding silently passing un-encoded text as is instead of barfing.
> (Eg,
> + ;; using these URLs in a browser and then copy-pasting it from the address
> + ;; should get you a properly encoded string instead.)
This parenthetical is not really correct. In particular, you cannot
rely on what's displaying in the browser address bar to be a
correctly-encoded URL -- the address bar is part of the user
interface, and not a reflection of the internal state.
For some discussion on this point, see this thread:
https://groups.google.com/d/topic/mozilla.dev.platform/V0XjJ-uQbcI/discussion
In general, the whole URL parser should probably be changed/re-written
to follow http://url.spec.whatwg.org/ but that's a longer-term
project.
Sam
_________________________
Racket Developers list:
http://lists.racket-lang.org/dev
