bug#11205: 24.0.95; starttls-use-gnutls no longer being honored

[nyc4bos]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Glenn Morris <rgm@gnu.org> writes:
>
>> You could always try redefining gnutls-available-p to return nil, eg:
>>
>> (defalias 'gnutls-available-p 'ignore)
>
> Yes.  In any case, this is neither a regression, nor a bug.  The
> variable in question controls what external program is used; not whether
> to use an external program or not.  So I'm closing this bug report.

The variable, `starttls-use-gnutls' docstring says:

"*Whether to use GnuTLS instead of the `starttls' command."

So if I'm understanding the docstring right, if it is nil, whatever is
the `starttls' command should be used, NOT GnuTLS.

[The defcustom really says:

(defcustom starttls-use-gnutls (not (executable-find starttls-program)))]

This appears to me to be a regression since setting `starttls-use-gnustls'
to nil still uses GnuTLS (when `starttls-program' is accessible), unlike
Emacs 23.4.

At a minimum, this is a documentation bug with regard to Emacs 24.

Can you explain to me why should a user now jump through more hoops
to use another program such as OpenSSL, in order to use TLS?

Setting `starttls-use-gnutls' to nil works just fine in Emacs 23.4

Thanks.