Re: [Extra] SEARCHRES / IMAP4rev2 clarifications

[Alexey Melnikov <[e-mail]>]

Hi Timo,

On 13/03/2020 14:09, Timo Sirainen wrote:
> On 13. Mar 2020, at 15.43, Alexey Melnikov <[e-mail] 
> <mailto:[e-mail]>> wrote:
>>
>>> 3) Interaction of CONTEXT and SEARCHRES:
>>>
>>> SEARCH RETURN (SAVE MIN PARTIAL 2:10) 1:100
>>>
>>> Is the $ value going to be 1 (MIN), 2:10 (PARTIAL), 1:10 
>>> (MIN+PARTIAL) or 1:100 (ALL)? I'm thinking 1:100 is the right answer 
>>> here.
>>
>> I think you found a real issue here. You are right that none of RFCs 
>> say anything about this case.
>>
>> I think my implementation is doing 2:10 (PARTIAL), but I need to 
>> double check. The intent of is for SAVE to use what is returned or 
>> what would have been returned. One can argue that "1:10 
>> (MIN+PARTIAL)" is also a valid choice here.
>>
>> I don't think setting $ to 1:100 (ALL) is right here, because this 
>> client only wants to know PARTIAL and MIN data. Data about ALL 
>> messages might not even be available in the server, depending on how 
>> PARTIAL is constructed.
>
> I guess this same question also applies to what happens even if MIN 
> isn't used. My current code is saving 1:100, but that's mainly because 
> of how the implementation happens to work right now (PARTIAL isn't 
> very optimized).
>
> I had also been thinking that maybe a client could work like:
>
> a SEARCH RETURN (SAVE PARTIAL 1:10) TEXT stuff
This means save the current partial segment into $, so I think it should 
be 1:10
> ..
> b SEARCH RETURN (PARTIAL 11:20) $
> ..
> c SEARCH RETURN (PARTIAL 21:30) $
>
> But I suppose that's not really something clients would do.
Probably not. And if they do, they better know what it means!
> Then again, if you do:
>
> a SEARCH RETURN (SAVE COUNT PARTIAL 1:10) 1:100
>
> COUNT would override PARTIAL here in any case, so $ would have to be 
> 1:100.
Yes. If this is surprising, then I think it should be clarified 
somewhere. If a client wants to retrieve COUNT, but save PARTIAL, it 
would have to do 2 commands.
> I wonder if we should just disallow SAVE+PARTIAL combination?

I would rather not. I think combination of SAVE and PARTIAL has a well 
specified semantics.

Best Regards,

Alexey