Reliability requires the capacity to execute a sequence of operations in a deterministic way, even under exceptional conditions. This allows you to ensure that resources are not leaked and that you can maintain state consistency without relying on application domain unloading (or worse, process restarts) to fix any corrupted state. Unfortunately, in the.NET Framework, not all exceptions are deterministic and synchronous, which makes it difficult to write code that is always deterministic in its ability to execute a predetermined sequence of operations. In this article Stephen Toub will show you why, and explore features of the .NET Framework 2.0 that help you to mitigate these situations and write more reliable code.
MSDN Magazine October 2005
View Complete Post