ibatis-user-cs mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexandre Grenier" <Alexandre.Gren...@markettools.com>
Subject RE: Noob questions...
Date Wed, 17 May 2006 19:07:12 GMT
Thanks Gilles, that's clears things up a lot. I was getting confused
between some of what is available in java and in .net. Glad to hear
groupby and 2.0 proxy features are coming up!


About my DBA, in this case the connection is actually fine. The problem
is the number of commands that get executed for a 3 level graph, my unit
test for a simpler aggregate does about 50. Seems all I can do now is
map a single select with 2 joins to a value object, and then compose the
graph using C#. I think with joins and groupby this could be done in a
single command. I'm not sure if there's a better solution. I'm thinking
using a repository [Fowler/Evans] pattern for each level would provide
both the graphing and lazy loading...


Before finding out about IBatis I had already written a decent generic
type cache... I'll read that article and see what I can come up with.


Thanks again!





From: Gilles Bayon [mailto:ibatis.net@gmail.com] 
Sent: Wednesday, May 17, 2006 2:32 PM
To: user-cs@ibatis.apache.org
Subject: Re: Noob questions...



On 5/17/06, Alexandre Grenier <Alexandre.Grenier@markettools.com> wrote:

	Hello All,


	I got the SVN IBatis going with .NET 2.0. I like it, and I have
some questions.


	- Where can I see a C# example of lazy loading / proxy,
preferably with a generic list...


I'm finishing to implement the proxy support for.NET  2.0


	- My app builds a 3 level object graph using properties with
select in resultMap, which ends up generating a LOT of commands (lots of
branches)... I've seen some stuff about "groupby" which would be really
useful, where can I find an example of that? Or anything else that would
help... I like the idea of having ibatis compose the tree for me. 


groupby is not currently support, expected after the next release.


	- Looking for pointers on using caching on a webfarm... how to
keep cache on 16 servers synchronized? Seems like flushOnExecute becomes
useless in that context. 


You must write your own cache manager based for example on
http://www.danga.com/memcached/ or cache object on a layer above the
data layer.


	- I read something about <sql> and <include> tags... is this a
Java only feature?

Yes, but you have same functionality with extends attribute on statement

	Our DBA tends to complain when there is more than 1 connection
to the database per http request... I don't feel that's realist, but any
pointers in reducing connections/queries as much as possible are

A call with iBATIS to retrieve a graph of object only use 1 connection
so I don't understand the complain of your DBA

	Other examples (links etc) on reducing the amount of reflection,
overhead and all are much appreciated. Maybe there's a best practices
guide available somewhere? 

The current implementation in SVN don't use any more reflection [by
default] :-) (you could revert it if you have problem)









View raw message