bug#25702: 26.0.50; `message-make-fqdn' break when `(system-name)' doesn't ends in `.local'

bug#25702: 26.0.50; `message-make-fqdn' break when `(system-name)' doesn't ends in `.local'

[Rasmus]

Hi,

The combination of 3485c1693ef (see #24570) and maybe the removal of
`message-valid-fqdn-regexp' means that `message-make-fqdn' has broken for
me.

My (system-name) is the model number of the machine I use (e.g. "X200s").
So now it's selected as the fdqn via `message-make-fqdn':

    ((not (string-match message-bogus-system-names sysname))
          ;; `system-name' returned the right result.
          sysname)

Thus, MESSAGE-ID are something@X200s, which is a bogus address.

For now I can add laptops model number to `message-bogus-system-names' but
that seems wrong.  Shouldn't user-domain be preferred over the sysname
anyway?

Thanks,
Rasmus

bug#25702: 26.0.50; `message-make-fqdn' break when `(system-name)' doesn't ends in `.local'

[Lars Ingebrigtsen]

Rasmus <rasmus@gmx.us> writes:

> The combination of 3485c1693ef (see #24570) and maybe the removal of
> `message-valid-fqdn-regexp' means that `message-make-fqdn' has broken for
> me.
>
> My (system-name) is the model number of the machine I use (e.g. "X200s").
> So now it's selected as the fdqn via `message-make-fqdn':
>
>     ((not (string-match message-bogus-system-names sysname))
>           ;; `system-name' returned the right result.
>           sysname)
>
> Thus, MESSAGE-ID are something@X200s, which is a bogus address.

I think this has been fixed in Emacs in 2017.  The code is now:

     ;; A system name without any dots is unlikely to be a good fully
     ;; qualified domain name.
     ((and (string-match "[.]" sysname)
	   (not (string-match message-bogus-system-names sysname)))
      ;; `system-name' returned the right result.
      sysname)

So it won't use sysname if there's no dot.

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