GNU bug report logs - #71295
29.3; url-retrieve-synchronously does not timeout if initial connection hangs

Previous Next

Full log

Message #11 received at 71295 <at> debbugs.gnu.org (full text, mbox):

From: "Aaron Zeng" <z <at> bcc32.com>
To: "Eli Zaretskii" <eliz <at> gnu.org>
Cc: 71295 <at> debbugs.gnu.org, azeng <at> janestreet.com
Subject: Re: bug#71295: 29.3;
 url-retrieve-synchronously does not timeout if initial
 connection hangs
Date: Mon, 03 Jun 2024 20:09:03 -0400

On Sat, Jun 1, 2024, at 15:38, Eli Zaretskii wrote:
> The timeout cannot work when Emacs is stuck in a system call trying to
> initiate a network connection.  That is blocking, unless you use
> url-retrieve asynchronously.  The timeout starts working after the
> connection is established, so it basically protects against too slow
> downloading of data after the initial connection.

That makes sense, although it's not what I would have guessed from the docstring.

> I think it's a bug in Spacemacs that it attempts to solve connectivity
> problems this way.

What would be a reasonable alternative?  This is the same approximate method used by, e.g., request.el.  See https://github.com/tkf/emacs-request/blob/01e338c335c07e4407239619e57361944a82cb8a/request.el#L767-L775.

Is the idea just to use url-retrieve directly?  I hacked together a quick function that doesn't seem to hang:

(defun url-retrieve-synchronously-but-dont-hang (url &optional silent inhibit-cookies timeout)
  (with-timeout (timeout
                 (error "Timed out retrieving %S after waiting for %s seconds"
                        url timeout))
    (let* (done
           (buf
            (url-retrieve url
                          (lambda (&rest _)
                            (setq done (current-buffer)))
                          nil silent inhibit-cookies)))
      (when buf
        (let ((proc (get-buffer-process buf)))
          (while (accept-process-output proc))
          done)))))

This is probably rather naive of me, but I guess now I'm wondering why url-retrieve-synchronously actually sets url-asynchronous to nil.  Is there a good reason not to use :nowait when it is available?  It seems like it would be useful to have a wrapper around url-retrieve that just "does what I mean" here.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.