DISQUS

Jonathan Pryor No. As miguel said:

Mono.Tasklets requires changes to the VM that are not portable to other ISO CLI implementations.

In short, This isn't pure C#, it requires runtime support.

Tigraine public virtual ISet<OrderItem> OrderItems
{
get { return orderItems_; }
set { orderItems_ = value; }
}

private ISet<OrderItem> orderItems_;

public Order()
{
orderItems_ = new HashedSet<OrderItem>();
}

I know it should work that way, but it doesn't.
Can't say why. But it doesn't work.

Maybe I should have added that the problem is occuring during my Repository unit tests. If I run the tests one by one it works, but if they get executed all together I get the exception.

I think I'll dedicate some more time to the problem tomorrow. But debugging concurrency issues sucks.

Tigraine That didn't work.. I tried having a private field that gets accessed through the property and that gets set by the ctor..
I also tried to set the field on demand from the property, but both times I got the "Illegal attempt to associate a collection with two open sessions".

Thanks anyway, I guess initializing the collection isn't all that difficult after all :)

robconery Weird cause they work on Chrome - which uses Webkit

robconery Hi Oren - if you implemented IRepository<T> as I've done here, how would this not work? Can you be specific in terms of "a lot of work" and what that means?

Justin Rudd yes and no. I built out a prototype in 2001 using POET. I was able to port over the OLTP portions in a few days because of good abstractions (and this was with C++). Now the OLAP stuff took weeks. Primarily because all the OLAP tools we built around POET had to be thrown away and rewritten from scratch. If I had to do it over again, I'd have a secondary system that would go through the POET DB every once and awhile and convert directly to a RDBMS data warehouse. But I was young and knew no better.

migueldeicaza Currently I emit new assemblies, doing LCG is on the to-do list;

Additionally, some folks have pointed out that getting a delegate back pointing to the generated code is useful, to avoid compiling the same code over and over (Ed Ropple), so that will also be done.

nstults Ahh Mr. Rahien, I do love how you sugarcoat everything you say. One day you'll walk into a blog and say just what you mean and shock everyone out of their boots.

I wasn't actually aware of BooParser.ParseString until just this moment. It looks to me, though, like the method gives you an AST tree, which you could then optionally compile and run. This is fine, but I'd still either need to compile the AST tree to my own object model or do a good bit of mucking around with the Boo AST to make sure custom variables were expanded properly, custom functions, etc. I am not sure it would have been any less work at the end of the day, and probably a bit more opaque.

That being said, I'm *not* sure Simple Expression Evaluator has a strong value proposition when compared to the two already existing Expression Evaluators on CodePlex, FLEE (uses dynamic code generation) and LazyParser (uses reflection). Both probably would work as well as mine, and existed before I wrote this. But I wanted a good excuse to try out Irony.net, and I selfishly wanted full control of the API. And, of course, I like this crap, and enjoy the tinkering (something I'm sure you could never understand :) ) Whether or not it provides anybody with a new and better expression eval library (probably not) it can serve as a pretty textbook example of one common approach to turning an AST into an object model and evaluating it.