As Colin explained, the issue originated from the implementation of M2M4RIA, which includes a dedicated code generator for generating m2m-specific client-side code. When you use m2m in your project by using the corresponding Nuget package, you may get into trouble if you're already using another code generator for Ria services. The problem is that in this case there is a) no way to specify how both code generators are to be combined, and b) there is (as far as I know) also now way the specify which code generator should be used. This may lead to unpredictable results, where your code may no longer compile or it may no longer work correctly.
It would be nice of the code generator process of Ria services was implemented as a customizable pipe-line where custom code generators could be easily plugged-in at specific locations. However, I'm aware of the fact that this would have a significant impact on the code generation process, because the extensible part is currently completely text-based. This makes it very hard, if not impossible to plug multiple code generators together.
Now that Ria services is open source and since we have plans to make m2m an integral part of Ria services, there will no longer be a need for a custom code generator for m2m itself. Still, it will be interesting to get some concrete ideas together about how such a pluggable code generator pipeline could be realized.
This content has not been rated yet.
6
Reputation
1
Total posts