Re: bug#13546: 24.2.92; Error(s) when sending emails

[Sebastien Vauban]

Eli,

Eli Zaretskii wrote:
>> From: "Sebastien Vauban" <wxhgmqzgwmuf@spammotel.com>
>>
>>>> And, what's even more surprising, is the second error: when wanting to
>>>> save a draft copy of the email (as I couldn't send it), with C-c C-d,
>>>> I've got the error:
>>>>
>>>> --8<---------------cut here---------------start------------->8---
>>>> basic-save-buffer-2: IO error writing d:/Users/sva/.gnus.d/drafts/drafts/2:
>>>> Socket operation on non-socket
>>>> --8<---------------cut here---------------end--------------->8---
>>>
>>> Is this all repeatable, or did it happen just once?
>>
>> FYI, the _first_ time I press `C-c C-d', I get the above. The second time,
>> not anymore, and the draft is correctly saved: I quit Emacs, restart it,
>> and send straight away my email.
>
> Sorry, I don't understand: is the "second time" in the same Emacs session,
> or after restarting?

Sorry for not being clear the first time ;-)

Let's recap:

1. The sending error occurs (and reoccurs again and again if I try to resend
   the email)
2. I'm trying to save the mail as a draft (C-c C-d): IO error above
3. I'm retring to save it once again ("second" time), with no other change:
   draft is successfully saved
4. I'm restarting Emacs
5. I'm sending the precited email

Note -- Looking at the location of the drafts, I'm puzzled that the draft is
saved in `~/.gnus.d/drafts' (that is, under my `gnus-directory'), and not in
`~/Mail/drafts' (that is, under my `message-directory'). I though that
everything related to "email composing" would be handled by Message?

> If the former, does the draft get saved to the same file
> d:/Users/sva/.gnus.d/drafts/drafts/2 as in the first attempt? Also, does
> "first time" mean first time after the laptop awakes from sleep mode? If so,
> can you wait a little longer before you try this, perhaps the disk did not
> yet have chance to spin up.

The problem occurs for example just now, while my laptop is up since this
midday. So, it's not 1just a few seconds or so after making the laptop alive
-- BTW, my new laptop (Asus Zenbook UX31) is 100% SSD, so I guess it can't be
more speedy than it is.

Why did I mention that fact, then?  Because, before, with the old laptop, I
quitted Emacs much more often -- at least once every evening. Now, I'm more or
less used to let Emacs (and the laptop) sleep between work sessions.

> In any case, I'm sure that the "Socket operation on non-socket" part is a
> red herring: write-region is not careful to zero out errno before the
> operations it wants to diagnose, so the value is probably coming from some
> entirely irrelevant interim error. What is important is that write-region
> _does_ fail in one of these operations, the question is which one?

I cannot reproduce yet the problem, as I still don't have a clue about the
conditions which are necessary to do so.

Any test I could do when this does happen?

Best regards,
  Seb

-- 
Sebastien Vauban