Querying related objects directly

Hi Alan,

I still don't know how to use #retrieve: but doing as you suggested
helped me reach the lowest level (ObjectD), but not ObjectE, because I
was wrong in my description, and ObjectD and ObjectE relation is not
1:1.

Putting everything in a the real context:

SurveyDefinition (ObjectA)
\_ 1:n CategorySurvey (ObjectB)
  \_1:n SurveySection (ObjectC)
    \_ 1:n SurveyQuestion (ObjectD)

In other hierarchy of equally composed objects, I have the "results"
for each SurveyDefinition.
SurveyResult (refs SurveyDefinition)
 \_1:n CategorySurveyResult (refs CategorySurvey)
   \_ 1:n SurveySectionResult (refs SurveySection)
     \_ 1:n QuestionAnswer (refs SurveyQuestion)

All the above described 1:n relations are bidirectional, it is, all
levels answer it's parent when receive #owner.
And I can send #survey to any of the described classes, and they will
delegate it's resolution until they reach an object with the direct
reference to SurveyDefinition, it is SurveyResult or CategorySurvey.

If I instantiate the whole result in memory, I can do whatever I like with them.

What I'm trying is to narrow the results, querying only certain
QuestionAnswer's according to some values, but only belonging to a
particular SurveyDefinition.

E.g.:
query := Query read: QuestionAnswer where: [:each | each value > 10
AND: [each survey = aSurveyDefinition]].

Because I'm only interested in a small subset of answers (<5%), I have
thousands of QuestionAnswers, there is no point in loading the
remaining 95% just to be discarded, with all the CPU and I/O time it
means.

Also SurveyQuestion is a whole hierarchy of classes, sharing a common
table, but using a FilteredTypeResolver.

I hope it is more clear now.


There is a quick "solution" for this that I'm trying to avoid:
I can add a survey_id field to QuestionAnswer's table, and declare a
OneToOneMapping in QuestionAnswer.
Maybe it's the best, and the fastest too. But I try to unify
references as much as I can.


Regards,

--
Esteban.


2014-10-16 17:48 GMT-03:00 Alan Knight <[hidden email]>:

> I don't have code here right now, but what happens if you do
> (Query read: ObjectA where: [:each | each = someA])
>    retrieve: [:a | a b c d e]
>
>
> On 16 October 2014 13:11, Esteban A. Maringolo <[hidden email]> wrote:
>>
>> Also: Worst mailing list subject ever. :)
>>
>> I have a complex, or not so, composed object relationship, and I want to
>> read elements of the lower level of the relation, filtering by the top
>> level.
>>
>> I have the following relations
>> ObjectA 1:n ObjectB
>> ObjectB 1:n ObjectC
>> ObjectC 1:n ObjectD
>>
>> ObjectE 1:1 ObjectD
>>
>> In the pristine ORM-free world, all objects know how to answer the
>> instance of ObjectA to which each one belongs/references.
>>
>> Ej,
>> ObjectE>>#objectA, ask for #objectA to anObjectD, and anObjectD ask for
>> its #objectA to its parent, and the same up to ObjectB, until you get the
>> answer which is the root of the whole tree, anObjectA.
>>
>> What I want is to perform a query reading instances of ObjectE filtering
>> only those "belonging" to ObjectA.
>>
>> Something like:
>> | query |
>> query := Query read: ObjectE where: [ :each | each objectA = anObjectA].
>>
>> In my ObjectE class model there is no reference nor mapping to #objectA,
>> nor there is a foreign key in ObjectE table, because to get to ObjectA table
>> you can joining all references together.
>>
>> I'm sure this is doable, I just don't know how to do it.
>>
>>
>> Best regards!
>>
>> --
>> Esteban.
>>
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "glorp-group" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> To post to this group, send email to [hidden email].
>> Visit this group at http://groups.google.com/group/glorp-group.
>> For more options, visit https://groups.google.com/d/optout.
>
>
> --

> You received this message because you are subscribed to a topic in the
> Google Groups "glorp-group" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/glorp-group/KRcO5pdt2Zw/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to

> [hidden email].
> To post to this group, send email to [hidden email].
> Visit this group at http://groups.google.com/group/glorp-group.
> For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "glorp-group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
To post to this group, send email to [hidden email].
Visit this group at http://groups.google.com/group/glorp-group.
For more options, visit https://groups.google.com/d/optout.

2014-10-20 0:35 GMT-03:00 Alan Knight <[hidden email]>:
> Yes, it's possible, but it does start to make the brain hurt. Or at least
> they make my brain hurt, having been away from the code for a while, and not
> even having a database running on this machine. So you can't really express
> a join that's got multiple clauses directly, but you can write an expression
> that does that. So I think it might work if you just express the join of a
> one to one mapping as [:each | each owner owner owner] and make it
> beForPseudoVariable. I think... But I don't see any examples of it in the
> tests, which is usually not a good sign for something working properly :-)

I'm coming back to this because being able to dynamically add a
condition to the where clause would save the retrieval of thousands of
useless objects.

What you mean when you say "express the join of a 1:1 mapping" using a
block (instead of a Join instance).
I can see Join's #asGlorpExpression returns self, and I get lost in
its polymorphism with GlorpExpression.

If I understood what you meant you say that in my class descriptor I
could do something similar to this (which doesn't works)

(aDescriptor newMapping: OneToOneMapping)
   attributeName: #survey;
   referenceClass: SurveyDefinition;
   join:  [:each | each owner owner owner];
   beForPseudoVariable

If something like this could be done it would be really powerful.

In my example case I know in advance the filtering condition
(#survey), but in the context of a dynamically defined filter (which
I'm working on), it would be great to translate those filters to
GlorpExpressions without having to define a column in my filtered
object, delegating everything to the already defined relations.

Regards!

Esteban A. Maringolo

--
You received this message because you are subscribed to the Google Groups "glorp-group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
To post to this group, send email to [hidden email].
Visit this group at http://groups.google.com/group/glorp-group.
For more options, visit https://groups.google.com/d/optout.

"degenerate expression" :)

Well... it seems I'm being too degenerated, because the following doesn't work.

Should the join be defined as follows?

(aDescriptor newMapping: OneToOneMapping)
   attributeName: #survey;
   referenceClass: SurveyDefinition;
   join:  [:each | each owner owner owner] asGlorpExpression;
   beForPseudoVariable


Esteban A. Maringolo


2014-10-20 19:38 GMT-03:00 Alan Knight <[hidden email]>:

> Yes, that's what I meant. A join is in some ways just a degenerate
> expression, and you should be able to use an arbitrary expression in its
> place.
>
>
>
> On 20 October 2014 14:56, Esteban A. Maringolo <[hidden email]> wrote:
>>
>> 2014-10-20 0:35 GMT-03:00 Alan Knight <[hidden email]>:
>> > Yes, it's possible, but it does start to make the brain hurt. Or at
>> > least
>> > they make my brain hurt, having been away from the code for a while, and
>> > not
>> > even having a database running on this machine. So you can't really
>> > express
>> > a join that's got multiple clauses directly, but you can write an
>> > expression
>> > that does that. So I think it might work if you just express the join of
>> > a
>> > one to one mapping as [:each | each owner owner owner] and make it
>> > beForPseudoVariable. I think... But I don't see any examples of it in
>> > the
>> > tests, which is usually not a good sign for something working properly
>> > :-)
>>
>> I'm coming back to this because being able to dynamically add a
>> condition to the where clause would save the retrieval of thousands of
>> useless objects.
>>
>> What you mean when you say "express the join of a 1:1 mapping" using a
>> block (instead of a Join instance).
>> I can see Join's #asGlorpExpression returns self, and I get lost in
>> its polymorphism with GlorpExpression.
>>
>> If I understood what you meant you say that in my class descriptor I
>> could do something similar to this (which doesn't works)
>>
>> (aDescriptor newMapping: OneToOneMapping)
>>    attributeName: #survey;
>>    referenceClass: SurveyDefinition;
>>    join:  [:each | each owner owner owner];
>>    beForPseudoVariable
>>
>> If something like this could be done it would be really powerful.
>>
>> In my example case I know in advance the filtering condition
>> (#survey), but in the context of a dynamically defined filter (which
>> I'm working on), it would be great to translate those filters to
>> GlorpExpressions without having to define a column in my filtered
>> object, delegating everything to the already defined relations.
>>
>> Regards!
>>
>> Esteban A. Maringolo
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "glorp-group" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> To post to this group, send email to [hidden email].
>> Visit this group at http://groups.google.com/group/glorp-group.
>> For more options, visit https://groups.google.com/d/optout.
>
>
> --

> You received this message because you are subscribed to a topic in the
> Google Groups "glorp-group" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/glorp-group/KRcO5pdt2Zw/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to

> [hidden email].
> To post to this group, send email to [hidden email].
> Visit this group at http://groups.google.com/group/glorp-group.
> For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "glorp-group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
To post to this group, send email to [hidden email].
Visit this group at http://groups.google.com/group/glorp-group.
For more options, visit https://groups.google.com/d/optout.