From: Roderick Bloem (roderick.bloem@colorado.edu)
Date: Wed Mar 28 2001 - 12:37:32 CST
Hi Daniel,
I agree that the headers do not comply to the standard. They should be
encoded. If I understand correctly, every & that does not separate
arguments, should be replaced by %26. I would argue that this is a bug
in hypermail.
However, I replaced the ampersands by hand in one mail message, and that
still does not cause netscape to copy the In-Reply-To identification.
As you point out, that is not a bug in netscape, but allowed behavior.
This does not solve one point, though: How come things worked under
Hypermail 2b29? Quite possibly, some of the mail in our archive was
replied to using a normal mailer. This would mean that the In-Reply-To
field was set, and was used by HM to match the thread. However, some of
the mail was replied to using the hypermail web interface, and it did
get matched correctly. This happens no longer. Did the subject-based
matching become stricter from 2b29 to 2b30?
With the matching as it currently is, the hypermail web interface loses
a lot of its usability. We used to just throw away messages that were
sent to the mailing list, because we could reply to them using the web
interface. That no longer works. Is there a work-around we can think
of? Maybe a form-based way to reply to a message, that would fill in
the correct In-Reply-To?
Roderick.
Daniel Stenberg wrote:
>
> On Fri, 23 Mar 2001, Roderick Bloem wrote:
>
> > If I take a message from the archive, and click [ respond to this message
> > ], that corresponds to the URL, say:
> >
> > "mailto:boldhsv@jobim.colorado.edu?Subject=Re:%20BUG%20FIXED:%20guided%20search%20for%20CTL&In-Reply-To=<3ABB843B.C8D4D127@colorado.edu>".
>
> This mailto: link does not conform to what RFC2368 (section 2) says about
> mailto: URLs:
>
> mailtoURL = "mailto:" [ to ] [ headers ]
> to = #mailbox
> headers = "?" header *( "&" header )
> header = hname "=" hvalue
> hname = *urlc
> hvalue = *urlc
>
> Note how multiple headers are separated. Then read a following paragraph from
> that same RFC:
>
> Because the "&" (ampersand) character is reserved in HTML, any mailto
> URL which contains an ampersand must be spelled differently in HTML
> than in other contexts. A mailto URL which appears in an HTML
> document must use "&" instead of "&".
>
> > Note that there is no In-Reply-To field!
>
> ... this might be because of this bad format. But also:
>
> The user agent interpreting a mailto URL SHOULD choose not to create
> a message if any of the headers are considered dangerous; it may also
> choose to create a message with only a subset of the headers given in
> the URL. Only the Subject, Keywords, and Body headers are believed
> to be both safe and useful.
>
> There's nothing that says it must use that header, even if it understands it.
>
> --
> Daniel Stenberg - http://daniel.haxx.se - +46-705-44 31 77
> ech`echo xiun|tr nu oc|sed 'sx\([sx]\)\([xoi]\)xo un\2\1 is xg'`ol