Re: draft-duerst-mailto-bis-06 - Detailed review of Get form submission and the To hvalue

Hello Michael,

Many thanks for your comments.

[I took the liberty to change the subject to -06.]

At 07:12 09/03/12, Michael A. Puls II wrote:
><http://tools.ietf.org/html/draft-duerst-mailto-bis-06>
>
>The following should submit "mailto:?subject=test" to the default mail client (whether that's Thunderbird, M2 or Outlook or Gmail etc.). The to=test value is dropped because GET submissions drop '?' and everything after it.
>
><form action="mailto:?to=test" method="GET">
>    <input name="subject" value="test">
>    <button>submit</button>
></form>

Yes.

>Now, with the following, you'll either get "mailto:test?subject=test" or "mailto:?subject=test" depending on how you interpret the mailto spec.
>
><form action="mailto:test" method="GET">
>    <input name="subject" value="test">
>    <button>submit</button>
></form>

Why either-or? Why the mailto spec? What happens with the form is
entirely an HTML matter. The 'action' attribute value of the <form>
element isn't the same as the 'submitted' URI. This applies not only
to the mailto: URI, but also to URIs of all other schemes. Implementation
of forms in a HTML user agent (read: browser) shouldn't have to know
about about specifics of any scheme at all.


>The mailto spec says that "mailto:test" is equivalent to "mailto:?to=test". If so, then you can see why an implementation might drop the 'test' value. But, you can also look at it the other way and say that according to generic form submission, the To value should stay.

Of course it should stay. Form submission should be generic.
It should be possible to use forms for a new "foobar:" scheme,
and have to change absolutely nothing in a browser, just have
it work like any other scheme.


>This means that 'mailto:test' and 'mailto:?to=test' are NOT equivalent as far as form sumbission goes.

This equivalence is relevant for resolution. Whether I hand over
'mailto:test' or 'mailto:?to=test' to a mailer should not matter.
For form submission, these two do not produce the same output
if used in the 'action' attribute of the <form> element.

Looking at the current draft, it seems that it could help to
move some text from Section 2, syntax, to Section, semantics.


>A note about this in the mailto spec (or some new mailto form submission spec) might help clear this up.

I'm still not sure why there's a need to "help clear this up".
Mentioning that the two above URIs don't result in the same
URI when processed according to HTML forms URI processing is
about the same as saying that when you use these two URIs
as input into an arbitrary program, you might get different
outputs as a result. There may be dozens of ways in which
to manipulate URIs.


><http://www.whatwg.org/specs/web-apps/current-work/multipage/forms.html#submit-mailto-headers> currently defines the 'mailto:' GET, PUT and DELETE submission as having different behavior between 'mailto:test' and 'mailto:?to=test'.

Does it do that specifically for the "mailto:" scheme?
Or does it do so generically? My expectation is that
it does so generically, but I'm currently traveling and
cannot look it up.


Regards,    Martin.


#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp

Received on Friday, 13 March 2009 03:33:41 UTC