Well, i just feel like that. Currently I'm trying to fix TRANSACT implementation I've been working on. This is by no means interesting for anybody but me. The thing is, implementation of such distributed programs are difficult no matter what. Further complications arise from my reluctance for implementing additional protocols, such as a separate uart for debugging and statistics.
This brings us to a general philosophical question. What is the best amount of generalization in any implementation. I stick to as general as possible rule. But I probably overdo :P. For instance, currently i am reporting using the same system as the experiments. This is not only limited also wierd since I need to decipher what is what. Dedicated implementations are better in this respect. They are specialized to some aspects.
My problem is just doing something means i need to change a lot of things when i change my mind, which is often. With a proper degree of abstraction, I also limit my options and hopefully I am more productive. Does this work for anyone else I have no idea. By the way, this kind of reflective writing looks really lame. Whatever. It is useful for creating content for a web page :)
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment