kill buffer created by url-retrieve-synchronously

kill buffer created by url-retrieve-synchronously

[Peter Münster]

Hi,

A timer calls periodically this function:

--8<---------------cut here---------------start------------->8---
(defun pm/get-json (url)
  "Copied from
http://frozenlock.org/2012/07/07/url-retrieve-and-json-api/."
  (let* ((url-mime-encoding-string "identity")
         (buffer (with-timeout (3 nil) (url-retrieve-synchronously url)))
         (result 
          (with-current-buffer buffer
            (goto-char (point-min))
            (re-search-forward "^$")
            (json-read))))
    (kill-buffer buffer)
    result))
--8<---------------cut here---------------end--------------->8---

Sometimes, the buffers created by url-retrieve-synchronously are not
killed, because I see them in (buffer-list).

What should I do to ensure, that these buffers are always killed at the
end of the function please?

TIA for any help,
-- 
           Peter

Re: kill buffer created by url-retrieve-synchronously

[Lars Magne Ingebrigtsen]

Peter Münster <pmlists@free.fr> writes:

> What should I do to ensure, that these buffers are always killed at the
> end of the function please?

Slap an `unwind-protect' around the main part of your function, and have
the `kill-buffer' in the unwind forms.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: kill buffer created by url-retrieve-synchronously

[Peter Münster]

On Sat, Feb 27 2016, Lars Magne Ingebrigtsen wrote:

> Slap an `unwind-protect' around the main part of your function, and have
> the `kill-buffer' in the unwind forms.

Thanks. But it still does not work. Just an idea: Is it possible, that
the with-timeout form kills the url-retrieve-synchronously function
because of the timeout, but the buffer has already been created?

Here my new function:

--8<---------------cut here---------------start------------->8---
(defun pm/get-json (url)
  "Copied from
http://frozenlock.org/2012/07/07/url-retrieve-and-json-api/."
  (let (my-json-buffer)
    (unwind-protect
        (let ((url-mime-encoding-string "identity"))
          (setq my-json-buffer (with-timeout (3 nil)
                                 (url-retrieve-synchronously url)))
          (with-current-buffer my-json-buffer
            (goto-char (point-min))
            (re-search-forward "^$")
            (json-read)))
      (if my-json-buffer
          (kill-buffer my-json-buffer)))))
--8<---------------cut here---------------end--------------->8---

-- 
           Peter

Re: kill buffer created by url-retrieve-synchronously

[Lars Magne Ingebrigtsen]

Peter Münster <pmlists@free.fr> writes:

> Thanks. But it still does not work. Just an idea: Is it possible, that
> the with-timeout form kills the url-retrieve-synchronously function
> because of the timeout, but the buffer has already been created?

Sorry; I didn't notice the `with-timeout' thing at all.

[...]

>         (let ((url-mime-encoding-string "identity"))
>           (setq my-json-buffer (with-timeout (3 nil)
>                                  (url-retrieve-synchronously url)))

[...]

>       (if my-json-buffer
>           (kill-buffer my-json-buffer)))))

Yes, the `with-timeout' is the reason the buffer isn't killed.  Because
it'll just return nil when timing out, and there's no way to get at the
buffer url.el created.

For reasons like this, `with-timeout' isn't recommended.
`url-retrieve-synchronously' should have a timeout parameter.  Feel free
to submit that as a wishlist bug report with `M-x report-emacs-bug'.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no