Re: Gnus Agent and mail

[Kevin Greiner <kgreiner@xpediantsolutions.com>]

Gaute B Strokkenes <gs234@cam.ac.uk> writes:

> On 12 mai 2003, kgreiner@xpediantsolutions.com wrote:
>>Gaute B Strokkenes <gs234@cam.ac.uk> writes:
>>>	* gnus-draft.el (gnus-group-send-queue): Bind
>>>	message-send-mail-real-function to the value of
>>>	gnus-agent-send-mail-function when sending queued mail.
>>
>>I don't like the idea of messing with
>>message-send-mail-real-function. What happens to someone who is not
>>using the agent (i.e. gnus-agent-send-mail-function is nil)?
>
> The sole purpose of that function is to send queued mail (it is the
> one that `J S' in the group buffer is bound to.)  So if you're not
> using the agent, you're not using this function.
>
>>What issue are you trying to resolve by adding this binding?
>
> Mail is sent by this function pretty much like any other mail.
> Therefore, before my patch it would be rewritten back to the queue
> folder if gnus were unplugged.  (Some people might like to be
> protected from hitting `J S' by mistake while unplugged, by I think
> it's a misfeature.  But I digress.)  With my patch, if
> gnus-agent-queue-mail is set to always, mail would always be rewritten
> to the queue folder if this hunk were not present.  Then there would
> be no way to send mail without temporarily setting
> gnus-agent-queue-mail to t or nil.

That last part was what I was hoping to here.  Wouldn't your patch
work just as well if you bound gnus-agent-queue-mail to nil as your
original bind of message-send-mail-real-function?  

> Hmm.  Actually, is is possible to enter the queue group and send
> messages with `D s' and `D S'.  I need to take care of those cases as
> well.

Hmmm... I'm glad you thought of that now.

Kevin