OK, I see them now. I'm writing up a little section on sharding and these
methods feel like they're not quite what I would want to use. Specifically,
they're using different criteria for insert and retrieval, which makes it
seem like things would get lost. In the example posting I referenced
earlier, you use the first character of the email address for insert but
then the primary key for retrieval. It seems like you would want to use the
email address for retrieval as well. Is there some magic going on in the
code that I'm missing? If not, then perhaps we need to morph this into a
trait on the field and use a MetaMapper def that points to that field as the
connection arbiter. Maybe something like:
trait ShardSelector[T, O <: Mapper[O]] extends MappedField[T,O] {
def selectConnection : PartialFunction[T,ConnectionIdentifier]
def connection = selectConnection(this.is)
}
And in Mapper:
val shardField : Can[ShardSelector] = Empty
Along with some appropriate hooks into the current code.
Then you could do
class MyUser extends Mapper[MyUser] {
object name extends MappedString(this) with ShardSelector {
def selectConnection = {
case n if n.charAt(0) < 'n' => dbAtoMConn
case _ => dbNtoZConn
}
}
override val shardField = Full(name)
}
Thoughts?
Derek
On Thu, Dec 4, 2008 at 4:17 PM, David Pollak
<[EMAIL PROTECTED]>wrote:
>
>
> On Thu, Dec 4, 2008 at 2:05 PM, Derek Chen-Becker <[EMAIL PROTECTED]>wrote:
>
>> OK, one last question for today before I start hacking at the text: You
>> mention some methods here for multiple databases with a single entity:
>
>
> On KeyedMetaMapper, override:
> def dbSelectDBConnectionForFind: PartialFunction[Type,
> ConnectionIdentifier] = Map.empty
>
> The ConnectionIdentifier defines which RDBMS to perform the query on.
>
> Also, on Mapper:
> def dbCalculateConnectionIdentifier: PartialFunction[A,
> ConnectionIdentifier] = Map.empty
>
>
>
>
>>
>>
>> http://groups.google.com/group/liftweb/msg/76c965c189a85103?pli=1
>>
>> But looking in the code I can't find those methods. Are they essentially
>> gone now? You also mentioned something better than sharding but I haven't
>> seen anything on the list (maybe I missed it).
>>
>> Thanks,
>>
>> Derek
>>
>>
>> On Thu, Dec 4, 2008 at 4:01 PM, Derek Chen-Becker <[EMAIL PROTECTED]>wrote:
>>
>>> Cool. I guess I'm editing my chapter this afternoon now :)
>>>
>>>
>>> On Thu, Dec 4, 2008 at 3:40 PM, David Pollak <
>>> [EMAIL PROTECTED]> wrote:
>>>
>>>>
>>>>
>>>> On Thu, Dec 4, 2008 at 7:33 AM, Derek Chen-Becker <
>>>> [EMAIL PROTECTED]> wrote:
>>>>
>>>>> I'm mostly done with the rough draft of our Record/Mapper chapter and I
>>>>> wanted to clarify a few points to make sure that my reading of the docs
>>>>> and
>>>>> code is accurate:
>>>>>
>>>>> 1. ByRef appears to me to essentially be an old-style join (using
>>>>> the where clause instead of join syntax). Is that accurate? Are there
>>>>> other
>>>>> use cases?
>>>>>
>>>>>
>>>> This allows you to compare two columns in the same row. Because mapper
>>>> does not allow for explicit joins or queries against more than one table at
>>>> once (with the exception of the In construct), there's no "old style join"
>>>> stuff.
>>>>
>>>> I've also changed the Join() QueryParam to PreCache... which is more
>>>> descriptive of what it does.
>>>>
>>>>
>>>>>
>>>>> 1.
>>>>> 2. It looks like the only way to get distinct results would be to
>>>>> use the findAllByPreparedStatement or findAllByInsecureSql methods. Is
>>>>> there
>>>>> some other way to do this that I'm missing?
>>>>>
>>>>> Or perhaps the Distrinct() QueryParam I just added. :-)
>>>>
>>>>
>>>>>
>>>>> 1.
>>>>> 2. The toForm method in Mapper has an overload that takes a
>>>>> redoSnippet. What is the use case for this?
>>>>>
>>>>> So you can keep the current state around over the course of multiple
>>>> form submissions when the validation fails.
>>>>
>>>>>
>>>>> 1. Will it be supported in Record or has that functionality been
>>>>> folded into something else?
>>>>>
>>>>> I hope it will make it into Record/Field.
>>>>
>>>>>
>>>>> 1.
>>>>> 2. It appears that the functionality of the buildSet... methods in
>>>>> MappedField has effectievly been replaced by the setFromAny in Field.
>>>>> Is
>>>>> this the case, or will those buildSet functions show back up in some
>>>>> subclass of field (it seems like there was a lot of overlap there)
>>>>>
>>>>> buildSet is JDBC specific and radically improves performance in
>>>> converting a JDBC column to something that can be put into a field.
>>>>
>>>>>
>>>>> 1.
>>>>> 2. Am I reading correctly that Record no longer retains the orginal
>>>>> value of a field when it's changed, but rather just flags the field as
>>>>> dirty? (there are no is/was methods on Field)
>>>>>
>>>>> Probably... but there will be is/was on Field. It's a very helpful
>>>> construct.
>>>>
>>>>>
>>>>> 1.
>>>>>
>>>>> These last two make sense to me since it seems like duplicated effort
>>>>> for #4 and not much of a use case on #5 that I can think of. Any info or
>>>>> comments would be appreciated!
>>>>
>>>>
>>>> Sure... wait a few minutes for my Distinct() checkin to make it to
>>>> GitHub
>>>>
>>>>
>>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Derek
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Lift, the simply functional web framework http://liftweb.net
>>>> Collaborative Task Management http://much4.us
>>>> Follow me: http://twitter.com/dpp
>>>> Git some: http://github.com/dpp
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
>
> --
> Lift, the simply functional web framework http://liftweb.net
> Collaborative Task Management http://much4.us
> Follow me: http://twitter.com/dpp
> Git some: http://github.com/dpp
>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Lift" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/liftweb?hl=en
-~----------~----~----~----~------~----~------~--~---
