Comments on: Adding Linq support to NPersist http://rogeralsing.com/2008/02/10/adding-linq-support-to-npersist/ Sat, 28 Jan 2012 14:35:30 +0000 hourly 1 http://wordpress.com/ By: Frans Bouma http://rogeralsing.com/2008/02/10/adding-linq-support-to-npersist/#comment-66 Sat, 23 Feb 2008 21:20:15 +0000 http://rogeralsing.wordpress.com/?p=55#comment-66 “Storing query related info in the context would be madness ”
I know about a given O/R mapper, something with Linq and Sql, which does that ;) :).

]]> By: Roger Alsing http://rogeralsing.com/2008/02/10/adding-linq-support-to-npersist/#comment-63 Sat, 23 Feb 2008 18:04:10 +0000 http://rogeralsing.wordpress.com/?p=55#comment-63 >>or are these spans stored inside the queryable produced by the Repository call and not inside the context?

Exactly.
Storing query related info in the context would be madness :-P

]]> By: Frans Bouma http://rogeralsing.com/2008/02/10/adding-linq-support-to-npersist/#comment-61 Sat, 23 Feb 2008 12:39:49 +0000 http://rogeralsing.wordpress.com/?p=55#comment-61 That prefetch path/span solution isn’t going to work if you formulate two queries using the same context, or are these spans stored inside the queryable produced by the Repository call and not inside the context?

I use a similar approach but with an extension method:
var q = (from c in metaData.Customer select c).WithPath(
new PathEdge(c=>c.Orders, o=>o.EmployeeId==3), …

the thing is with this you can specify nested edges on the fly and also specify filters with the edge.

One disadvantage is that if this query is inside a join branch, the whole path doesn’t make sense. This is the fault of the developer so ignoring it or throwing an exception is best.

It’s impressive how fast you got this working, I struggled for months with the provider, as there was no documentation about anything and I couldn’t pass on the elements in a query which looked very similar to the linq query, unfortunately.

Once a core system of a linq provider is in place the rest is filling in the blanks but don’t underestimate the fine print. There is a lot of time to waste on tiny little details, like all the different query forms ‘Contains’ can be used in, which results in nested aggregates and a lot of derived tables.

]]>