Re: No-fail `document` Function Spec

* John L. Clark <[email protected]> [2006-02-28 22:40]:
>On Tue, Feb 28, 2006 at 10:07:34PM +0100, A. Pagaltzis wrote:
>> * John L. Clark <[email protected]> [1970-01-01 01:00]:
>Whoa, how did this happen? -----------^^^^^^^^^^^^^^^^
>Bizarre.  Anyway.

That happened because I'm replying to a mail from the gzipped
mbox available from the archives, which puts a datetime format in
the `Date` header that mutt can’t parse.

In case someone else ever needs to fix those up, the following
oneliner will do the trick:

    perl -MHTTP::Date -pi -e's/^(Date:\s*)(.*)/"$1".time2str str2time $2/e' mboxfile

>For example, I try to pull down <http://www.w3.org/Member/>
>using `res:document`, and I get back an http protocol error
>report with 401 in the code field.  If I had access to the
>`WWW-Authenticate` header information (in this case, `Basic
>realm="W3C-Member"', likely shunted into a more pleasant XML
>structure), what would I do with it?

In case of Basic auth, I guess not much. OTOH with Digest auth,
the first request always yields 401 because you need a nonce from
the server before you can authenticate yourself, and that nonce
comes back in the `WWW-Authenticate` header of the 401 response.

Hmm, although you can’t set headers on the request anyway, and
the Digest auth C-R roundtrips would be handled by the HTTP
library beneath the function anyway.

Err, okay. I missed most of the picture. What I was proposing
would make sense to have in a more fully-featured extension for
doing HTTP, not this function. Ignore me.

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>