bug#35969: 26.2, Excorporate

[Thomas Fitzsimmons]

Hi,

"tenspd137 ." <dcday137@gmail.com> writes:

> So - I am not sure if I did it correctly, but I copied this function
> with Andreas' changes into a file:
>
> (defun url-https-proxy-after-change-function (_st _nd _length)
>   (let* ((process-buffer (current-buffer))
>          (proc (get-buffer-process process-buffer)))
>     (goto-char (point-min))
>     (when (re-search-forward "^\r?\n" nil t)
>       (backward-char 1)
>       ;; Saw the end of the headers
>       (setq url-http-end-of-headers (set-marker (make-marker) (point)))
>       (url-http-parse-response)
>       (cond
>        ((null url-http-response-status)
>         ;; We got back a headerless malformed response from the
>         ;; server.
>         (url-http-activate-callback)
>         (error "Malformed response from proxy, fail!"))
>        ((= url-http-response-status 200)
>         (if (gnutls-available-p)
>             (condition-case e
>                 (let ((tls-connection (gnutls-negotiate
>                                        :process proc
>                                        :hostname (url-host url-current-object)
>                                        :verify-error nil)))
>                   ;; check certificate validity
>                   (setq tls-connection
>                         (nsm-verify-connection tls-connection
>                                                (url-host url-current-object)
>                                                (url-port url-current-object)))
>                   (with-current-buffer process-buffer (erase-buffer))
>                   (set-process-buffer tls-connection process-buffer)
>                   (setq url-http-after-change-function
>                         'url-http-wait-for-headers-change-function)
>                   (set-process-filter tls-connection 'url-http-generic-filter)
>                   (process-send-string tls-connection
>                                        ;; Use the non-proxy form of the 
> request
>                                        (let (url-http-proxy)
>                                          (url-http-create-request))))
>               (gnutls-error
>                (url-http-activate-callback)
>                (error "gnutls-error: %s" e))
>               (error
>                (url-http-activate-callback)
>                (error "error: %s" e)))
>           (error "error: gnutls support needed!")))
>        (t
>         (url-http-debug "error response: %d" url-http-response-status)
>         (url-http-activate-callback))))))
>
> and then loaded it before running excorporate.  After that, I did M-x
> excorporate, and the minibuffer returns:  error in process filter:
> Server response is not an XML document

In this scenario, if you immediately (without restarting Emacs/reloading
anything) re-run M-x excorporate does it still fail?  I just want to
make sure that's not a transient failure.  If it does fail the second
time, can you post the HTTP response from the server?

> When I do the similar test by loading the url-http-create-request with
> Thomas's changes, I can get a connection and grab my schedule
> through the proxy.

OK.

> Let me know if I need to try something different.

Are you in a position to build Emacs master tip and retry the experiment
without patching anything?

Thanks,
Thomas

> On Mon, Jun 17, 2019 at 4:08 PM Thomas Fitzsimmons <fitzsim@fitzsim.org> 
> wrote:
>>
>> Hi,
>>
>> Good to hear that the patch I posted worked!
>>
>> Yes, that's the patch that Andreas's commit
>> 84613dae5c34ea742dd9a3e56f5acb55f604b483 applied.  From what I can tell,
>> you will not have that in Emacs 26.2.
>>
>> Can you try reverting my patch and applying Andreas's patch, and see if
>> Excorporate still works through the proxy?
>>
>> Thanks,
>> Thomas
>>
>> "tenspd137 ." <dcday137@gmail.com> writes:
>>
>> > The patch Thomas seems to work from behind the proxy.  My current
>> > emacs version is 26.2, so I would think it would include the commit
>> > Andreas is talking about....  I went and looked it up - is this the
>> > correct commit?
>> >
>> > diff --git a/lisp/url/url-http.el b/lisp/url/url-http.el
>> > index 53798f7..817c5ce 100644
>> > --- a/lisp/url/url-http.el
>> > +++ b/lisp/url/url-http.el
>> > @@ -1412,7 +1412,9 @@ The return value of this function is the
>> > retrieval buffer."
>> > 'url-http-wait-for-headers-change-function)
>> > (set-process-filter tls-connection 'url-http-generic-filter)
>> > (process-send-string tls-connection
>> > - (url-http-create-request)))
>> > + ;; Use the non-proxy form of the request
>> > + (let (url-http-proxy)
>> > + (url-http-create-request))))
>> > (gnutls-error
>> > (url-http-activate-callback)
>> > (error "gnutls-error: %s" e))
>> >
>> > Thanks!
>> >
>> > -C
>> >
>> >
>> > On Sat, Jun 15, 2019 at 1:41 AM Andreas Schwab <schwab@linux-m68k.org> 
>> > wrote:
>> >>
>> >> On Jun 14 2019, Thomas Fitzsimmons <fitzsim@fitzsim.org> wrote:
>> >>
>> >> > diff --git a/lisp/url/url-http.el b/lisp/url/url-http.el
>> >> > index 00803a103a..723d111d58 100644
>> >> > --- a/lisp/url/url-http.el
>> >> > +++ b/lisp/url/url-http.el
>> >> > @@ -329,7 +329,10 @@ url-http-create-request
>> >> >               ;; The request
>> >> >               (or url-http-method "GET") " "
>> >> >               (url-http--encode-string
>> >> > -              (if using-proxy (url-recreate-url url-http-target-url) 
>> >> > real-fname))
>> >> > +              (if (and using-proxy
>> >> > +                       (not (equal "https" (url-type 
>> >> > url-http-target-url))))
>> >> > +                  (url-recreate-url url-http-target-url)
>> >> > +                real-fname))
>> >>
>> >> That should already be handled by commit 84613dae5c.
>> >>
>> >> Andreas.
>> >>
>> >> --
>> >> Andreas Schwab, schwab@linux-m68k.org
>> >> GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
>> >> "And now for something completely different."