Re: [O] Possible bug: Emacs hangs on (require 'org)

[Eric Abrahamsen]

Luke <address@hidden> writes:

> On 23/11/16 10:39, Eric Abrahamsen wrote:
>> Luke <address@hidden> writes:

[...]

>>> I tried creating a minimal-org.el file with the following contents:
>>>
>>>     ;; activate debugging
>>>     (setq debug-on-error t
>>>           debug-on-signal nil
>>>           debug-on-quit nil)
>>>
>>>     ;; add latest org-mode to load path
>>>     (add-to-list 'load-path (expand-file-name "~/elisp/org-mode/lisp"))
>>>     (add-to-list 'load-path (expand-file-name
>>> "~/elisp/org-mode/contrib/lisp" t))
>>>
>>> And then running:
>>>
>>>     $ emacs -Q -l minimal-org.el
>>>
>>> ...but I still get the same result (Emacs hangs when I run `M-x
>>> (require 'org)`).
>>>
>>> Can anyone help tell me what's wrong, or suggest how I try and track
>>> down the problem?
>>
>> You're setting debug-on-quit to nil, but it's likely that setting it to
>> t is exactly what you need. Toggle it to t, eval (require 'org), then
>> while it's hung hit C-g. You should get a backtrace showing you what
>> Emacs was actually doing when you quit.
>>
>> Eric
>>
>>
>>
>
> Thanks to Eric for the debugging tip. Below is the backtrace that was
> produced. It looks like tramp is trying to make some connection to the
> host-name "host.does.not.exist" (which will obviously fail).
>
> This makes sense, because when I disable networking then Emacs starts
> immediately and doesn't hang. I'm not sure how to fix this though.

I don't have a copy of Emacs 24 handy (and I recommend you upgrade to
25), but as a band-aid you should be able to kill the check by setting
the `tramp-use-ssh-controlmaster-options' option to nil. That ought to
prevent tramp from testing the connection at all (at least, I think it
will).

In 25 it looks like "host.does.not.exist" has been replaced with a bogus
IP address, meaning that you won't time out on a DNS lookup, which is
likely what's hanging for you.

As for the real root of the problem -- why loading org results in tramp
doing this check -- I unfortunately can't tell you!

The real "fix", I think, would be upgrading to Emacs 25.

Eric