[Bug 32125] Should support auth using captcha

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=32125

--- Comment #41 from Dario Freddi <drf54321 at gmail.com> 2012-02-14 03:12:35 PST ---
(In reply to comment #40)
> (In reply to comment #38)
> > * The Label for any captcha except "qa" may be ''.
> 
> I've added this to my branch and updated the HTML. Review, please?

+            other than <code>qa</code>, connection managers SHOULD set Label
+            to an empty string instead of generating their own text.

What if protocol specifies the label anyway even if the challenge is, say, OCR? And most of all, what if a friendly proprietary protocol implements the captcha
type "lotsofcrap" which actually requires a label? I'd rephrase that part so that label should be set to an empty string if the protocol does not supply one.

Or actually, if you really intend to always provide the label from the handler unless it's part of the challenge (which makes sense), wouldn't it make sense to
ditch the label property in the structure, and provide the label in the payload instead? Although this is not matching the XMPP spec 1:1 it seems to me the
most flexible approach for handlers. In this case, qa will simply provide a payload consisting of plain text.

Now that I wrote this down in my stream of consciousness, I think I am strongly favorable towards ditching the label. After all, it doesn't make much sense to
have a field just for the sake of matching the spec (also, mainly for a captcha type we're not likely to be ever handling). Also, providing the label in the
payload prevents confusion from the handler side, which is much better off in providing a "Please answer this question: <payload>". We can advertise the
mimetype to be plain text in that case, and it will all quite fit together.

What do you reckon?

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA Contact for the bug.