Re: Repositories Global only or contextual?

The whole infinite scrolling idea technical - in a massively distributed
system such as twitter/facebook, it's very hard (expensive) to provide an
exact size of a list.  Plus, for most users the list is effectively
infinite..

Paging, and to a lesser extent filtering on the set-based repository
methods is part of the domain.  You're modelling "give me all albums by the
Rolling Stones, but do it 10 at a time because I can't carry all of them in
one go".

Of course you should try to implement this efficiently, and in a way that
doesn't distract from the semantics of the domain.  This often means
deciding between passing expressions/query objects (which can handle
AND/OR/NOT) or simple constraint lists (which can handle AND).

For example, I've seen people implement filtering on the UI which sent a
string expression tree to the webserver, which parsed the expression string
into an expression tree which was compiled to SQL in the Repository
implementation (well, DAL, they weren't using the repository patter).  That
worked well, the conversions were handled by framework infrastructure so
the domain code wasn't cluttered.  But it obviously took a lot of work to
get right...

-- 
Ryan


On 27 January 2014 21:22, Caleb Cushing <[email protected]> wrote:

>
>
> I've been dead for the last week with a sickness. But thanks all for the
>  input.
>
> On Fri, Jan 24, 2014 at 2:10 PM, <[email protected]> wrote:
>
>> But most repositories suffer from infrastructure leakage - take most
>> basic feature of paging. Paging will require additional arguments on your
>> methods or type signature suited for such operations. Neither of those is
>> part of your domain, but you need to have it or you will have performance
>> issues.
>>
>> Actually, I'm not entirely convinced on some reflection that Paging isn't
> part of the Domain. One of the principles of DDD is to ask, if you remove
> the computer, how is it done? take books, if I remove the computer would I
> print things on a scroll and unroll it.... No, because people don't
> actually like to consume things that way, it's too much information. We
> tend to think of this as a machine perfomance issue only, or a screen
> realestate issue, but I think there's truth that we as people don't want to
> consume an infinite list. I've found a few of the sites that provide said
> infinite scrolling to have rather frustrating user interfaces (personal
> opinion). When the list size is known, and it's seems humanly consumable
> it's not so bad. When reviewing a checkbook balance log, we eventually have
> to turn the page. I'm not 100% on this, but I think we may to some extent
> assume that the "technology" was the reason for this, but I think the human
> factor is the reason we stopped using scrolls, and humans are a domain
> concern (IMHO).
>
>
> --
> Caleb Cushing
>
> http://xenoterracide.com
>
> Calendar:
>
> https://www.google.com/calendar/embed?src=xenoterracide%40gmail.com&ctz=America/Chicago
>
>
> 
>