Re: [PATCH v2 06/10] cli: Introduce "notmuch address" command

[Mark Walters <markwalters1009@gmail.com>]

On Wed, 05 Nov 2014, Michal Sojka <sojkam1@fel.cvut.cz> wrote:
> On Wed, Nov 05 2014, Mark Walters wrote:
>> On Tue, 04 Nov 2014, Michal Sojka <sojkam1@fel.cvut.cz> wrote:
>>> On Tue, Nov 04 2014, Mark Walters wrote:
>>>> On Mon, 03 Nov 2014, Michal Sojka <sojkam1@fel.cvut.cz> wrote:
>>>>> This moves address-related functionality from search command to the
>>>>> new address command. The implementation shares almost all code and
>>>>> some command line options.
>>>>>
>>>>> Options --offset and --limit were intentionally not included in the
>>>>> address command, because they refer to messages numbers, which users
>>>>> do not see in the output. This could confuse users because, for
>>>>> example, they could see more addresses in the output that what was
>>>>> specified with --limit. This functionality can be correctly
>>>>> reimplemented for addresses later.
>>>>
>>>> I am not sure about this: we already have this anomaly for output=files
>>>> say. Also I can imagine calling notmuch address --limit=1000 ... to get
>>>> a bunch of recent addresses quickly and I really am wanting to look at
>>>> 1000 messages, not collect 1000 addresses.
>>>
>>> I think that one of the reasons for having the new "address" command is
>>> to have cleaner user interface. And including "anomalies" doesn't sound
>>> like a way to achieve this. I think that now you can use "date:" query
>>> to limit the search.
>>>
>>> I volunteer to implement "address --limit" properly after 0.19. This
>>> should be easy.
>>
>> I think this depends on how you view limit: is it to limit the output
>> (roughly to run "head" on the output), or is to bound the amount of work
>> notmuch has to do (eg to make sure you don't get a long delay). Your
>> suggestion is definitely the former, whereas I am more worried about the
>> latter: limit in your definition could take an essentially unbounded
>> amount of time.
>
> Why? If I understand you correctly, you think of limit in terms of
> messages. There is 1:N mapping between messages and addresses, where
> N >= 1. If I limit the number of printed addresses, I limit the number
> of messages as well. Only if N is zero (which probably can be the case
> with Bcc and --output=recipients) then it can result in unbounded work
> (provided you have infinite number of Bcc only messages in your
> database :-)).

Hi 

I was assuming the limit in your scheme would come after the
deduplication: so notmuch would have to find "limit" distinct
addresses. If the limit is applied before the deduping then I agree work
is bounded (in any sane case).

If limit is applied before the deduping then that seems fine.

Best wishes 

Mark

>
> Do I miss something?
>
> -Michal