Re: HTTP redirects make url-retrieve-synchronously asynchronous

[Mark Plaksin]

Stefan Monnier <address@hidden> writes:

>>>> url-retrieve-synchronously becomes asynchronous when HTTP redirects are
>>>> involved.  When it encounters a redirect, url-http-parse-headers calls
>>>> url-retrieve instead of url-retrieve-synchronously.  Naively switching to
>>>> the latter doesn't solve the problem and I haven't been able to find a 
>>>> fix..
>>> I believe I've fixed the problem with the patch below.
>> Thanks, but this doesn't fix it for me.
>
> Hmm... it does for me.  Have you recompiled the file?

Yes.  I even ran one more 'cvs update; make maintainer-clean; make
bootstrap'.  I must be missing something simple though because it still
fails for me.

>> In my original post I didn't make it completely clear that you have to set
>> mm-url-use-external to nil.
>
> That was obvious enough, otherwise the URL library isn't used.
> I actually reproduced your problem while directly calling
> url-retrieve-synchronously and the patch fixed the problem for me.

For me, both before and after the patch, url-retrieve-synchronously creates
two buffers: *http slashdot.org:80* and *http rss.slashdot.org:80*.  It
returns the former.  The slashdot.org buffer contains the redirect and the
rss.slashdot.org buffer contains the contents of their RSS feed.

When I run mm-url-insert, it returns ("http://slashdot.org/index.rss"; 316)
and inserts the redirect message into the current buffer (*scratch* for
me) instead of the contents of the feed.