On Friday, Mar 29, 2002, at 10:46 US/Pacific, Mark Crispin wrote:

> On Fri, 29 Mar 2002 10:22:14 -0800, Grant Baillie wrote:
>> Well, I understand that they need to change sometimes. But leaving 
>> open the option of their changing any time, without notice, seems 
>> wrong.
>
> As indeed it is.  But it isn't "any time".  There are only two points 
> at which capabilities may change: switching on TLS (STARTTLS command), 
> and authentication (AUTHENTICATE and LOGIN command).
>
>> Otherwise, how can clients be sure that it's OK to cache them? In 
>> principle (if not in practice), they'd have to issue CAPABILITY before 
>> issuing any extension command, for example.
>
> No.
>
> If you think about it, capabilities can't change once the client has 
> logged in, and you can rely upon that.

I agree that this is the logical and desirable behaviour, but it doesn't 
seem to be spelled out clearly in the spec. Did I miss something? Or is 
it just supposed to be something you're supposed to infer?

________________________________________
Grant Baillie
Mac OS X Mail - Apple Computer, Inc.
________________________________________

Reply via email to