changed default current-alist-separator-mode to amp, added semi-or-amp mode

svn: r7057

original commit: 0e2a1a465ac5bb8d0e5a67ed115e584e0faef1e9

collects/net/uri-codec-unit.ss

@@ -1,97 +1,15 @@
-;; 1/2/2006: Added a mapping for uri path segments
-;; that allows more characters to remain decoded
-;; -robby
-
-
 #|
 
-People often seem to wonder why semicolons are the default in this code,
+People used to wonder why semicolons were the default.  We then
-and not ampersands. Here's are the best answers we have:
+decided to switch the default back to ampersands --
 
-From: Doug Orleans <dougorleans@gmail.com>
+  http://www.w3.org/TR/html401/appendix/notes.html#h-B.2.2
-To: plt-scheme@list.cs.brown.edu
-Subject: Re: [plt-scheme] Problem fetching a URL
-Date: Wed, 11 Oct 2006 16:18:40 -0400
-X-Mailer: VM 7.19 under 21.4 (patch 19) "Constant Variable" XEmacs Lucid
 
-Robby Findler writes:
+  We recommend that HTTP server implementors, and in particular, CGI
- > Do you (or does anyone else) have a reference to an rfc or similar that
+  implementors support the use of ";" in place of "&" to save authors
- > actually says what the syntax for queries is supposed to be?
+  the trouble of escaping "&" characters in this manner.
- > rfc3986.txt (the latest url syntax rfc I know of) doesn't seem to say.
 
-The HTML 4.01 spec defines the MIME type application/x-www-form-urlencoded:
+See more in PR8831.
-
-http://www.w3.org/TR/html401/interact/forms.html#form-content-type
-
-See also XForms, which uses a "separator" attribute whose default
-value is a semicolon:
-
-http://www.w3.org/TR/xforms/slice11.html#serialize-urlencode
-http://www.w3.org/TR/xforms/slice3.html#structure-model-submission
-
---dougorleans@gmail.com
-_________________________________________________
-  For list-related administrative tasks:
-  http://list.cs.brown.edu/mailman/listinfo/plt-scheme
-
-
-
-
-
-
-From: John David Stone <stone@math.grinnell.edu>
-To: plt-scheme@list.cs.brown.edu
-Subject: Re: [plt-scheme] Problem fetching a URL
-Date: Wed, 11 Oct 2006 11:36:14 -0500
-X-Mailer: VM 7.19 under Emacs 21.4.1
-
------BEGIN PGP SIGNED MESSAGE-----
-Hash: SHA1
-
-        Danny Yoo:
-
- > > Just out of curiosity, why is current-alist-separator-mode using
- > > semicolons by default rather than ampersands?  I understand that
- > > flexibility is nice, but this is the fifth time I've seen people hit this
- > > as a roadblock; shouldn't the default be what's most commonly used?
-
-        Robby Findler:
-
- > It is my understanding that semi-colons are more standards compliant.
- > That's why it is the default.
-
-        According to the RFC1738 (http://www.ietf.org/rfc/rfc1738.txt),
-semicolons and ampersands are equally acceptable in URLs that begin with
-`http://' (section 5, page 17), but semicolons are ``reserved'' characters
-(section 3.3, page 8) and so are allowed to appear unencoded in the URL
-(section 2.2, page 3), whereas ampersands are not reserved in such URLs and
-so must be encoded (as, say, &#38;).  RFC2141
-(http://www.ietf.org/rfc/rfc2141.txt) extends this rule to Uniform Resource
-Names generally, classifying ampersands as ``excluded'' characters that
-must be encoded whenever used in URNs (section 2.4, page 3).
-
-        The explanation and rationale is given in Appendix B
-(``Performance, implementation, and design notes,''
-http://www.w3.org/TR/html401/appendix/notes.html) of the World Wide Web
-Consortium's technical report defining HTML 4.01.  Section B.2.2
-(``Ampersands in URI attribute values'') of that appendix notes that the
-use of ampersands in URLs to carry information derived from forms is, in
-practice, a serious glitch and source of errors, since it tempts careless
-implementers and authors of HTML documents to insert those ampersands in
-URLs without encoding them.  This practice conflicts with the simple and
-otherwise standard convention, derived from SGML, that an ampersand is
-always the opening delimiter of a character entity reference.  So the World
-Wide Web Consortium encourages implementers to use and recognize semicolons
-rather than ampersands in URLs that carry information derived from forms.
-
------BEGIN PGP SIGNATURE-----
-Version: GnuPG v1.4.5 (GNU/Linux)
-Comment: Processed by Mailcrypt 3.5.8+ <http://mailcrypt.sourceforge.net/>
-
-iD4DBQFFLR16bBGsCPR0ElQRAizVAJddgT63LKc6UWqRyHh57aqWjSXGAJ4wyseS
-JALQefhDMCATcl2/bZL0bw==
-=W2uS
------END PGP SIGNATURE-----
 
 |#
 
@@ -305,7 +223,8 @@ JALQefhDMCATcl2/bZL0bw==
   ;; http://www.w3.org/TR/html401/appendix/notes.html#ampersands-in-uris
   ;; listof (cons symbol string) -> string
   (define (alist->form-urlencoded args)
-    (let* ([sep (if (eq? (current-alist-separator-mode) 'amp) "&" ";")]
+    (let* ([sep (if (memq (current-alist-separator-mode) '(semi semi-or-amp))
+                  ";" "&")]
            [format-one
             (lambda (arg)
               (let* ([name  (car arg)]
@@ -342,9 +261,9 @@ JALQefhDMCATcl2/bZL0bw==
   (define current-alist-separator-mode
     (make-parameter 'amp-or-semi
       (lambda (s)
-                      (unless (memq s '(amp semi amp-or-semi))
+        (unless (memq s '(amp semi amp-or-semi semi-or-amp))
           (raise-type-error 'current-alist-separator-mode
-                                          "'amp, 'semi, or 'amp-or-semi"
+                            "'amp, 'semi, 'amp-or-semi, or 'semi-or-amp"
                             s))
         s))))