Comments on: The Software Practitioner’s Digest: October 2011

By: davidk

davidk — Thu, 17 Nov 2011 19:32:18 +0000

@Gustavo Thanks to the reference to the EUROMICRO paper. Although I only skimmed the paper it seems to me looser-coupling is the main difference. I take away that SOSE is a subset of CBSE with some design flexibility removed by sticking to the same choice for certain thing (like location independence). If SOSE is a ‘good’ paradigm then the removed flexibility was in areas where people frequently made poor choices or always he same choice. Hopefully we’ve learned something from CBSE and taken the best parts/practices forwards to make SOSE. [Although personally I think it’s just a new name for making the obviously correct choices when building software.]

By: Gustavo

Gustavo — Tue, 15 Nov 2011 21:49:52 +0000

@davidk, you're correct in that the differentiation I offered is is limited, hence I started by saying that my explanation was inaccurate. I was afraid to diverge too much by offering a more accurate explanation and I thought that binding time was the most significant difference. In the EUROMICRO paper I mentioned above, the authors point out that late binding is also possible in CBSE and cite COM as an example. They cover other differences which, in my opinion, are not as significant as binding time. Looked from another perspective, services in SOSE are black boxes for their clients, while CBSE allows components to be used as white, gray or black boxes. I'd imagine this distinction can be useful in areas like critical systems development, where redundancy and diversity strategies like NVP are used and SOSE could potentially be deemed the only acceptable paradigm.

By: davidk

davidk — Tue, 15 Nov 2011 15:48:10 +0000

Your SOSE vs CBSE differentiation is only describing late vs early binding. Component based systems often use late-binding (ex. plug-ins). So either there is something else important about SOSE or folks are just recycling the same old ideas to write new papers.