Hello Everyone. Welcome to my blog about my PhD journey. Its already been around 28 days since the journey started when I flew out from Auckland. Many interesting things have happened since then. I will be making many posts in the near future to fill in the gaps for those 28 days. I have been postponing making posts on the blog since I still dont have my own computer (I should hopefully have one in the next couple of days), but I have now been sort of pushed into starting off this blog because one of the courses that I have taken up requires maintaining a blog. The rest of this post will be about my views about the introductory lecture of that course. All of you Kiwiplan and Software Engineering buddies might be interested in reading the rest of this post. For the rest, just give it a try anyways! :-)
So, the name of the course is Advanced Topics In Software Engineering (CS 527) and it is taught by none other than Prof Ralph Johnson who is one of the Gang of Four who authored Design Patterns: Elements of Reusable Object-Oriented Software. The introductory lecture was an introduction to Patterns and Software Architecture in general.
During the first lecture, there were some references to a parallel between building architecture and software architecture. That is something that I have always wondered about. Will software architects ever be able to comprehensively specify the architecture of a software system and be sure that the specified system will definitely work, in the same way as building architects specify building plans? I guess that is a debatable topic. Some people may say that software architects can already achieve that and others may say that software architects should not even try to emulate building architects since the two areas are not comparable!
7 comments:
This is Hugh.
I believe that SE Arch can only be partially evaluated to work as they should have. This is due the lack of adequate architecture modelling tools that SE people could use. Civil engineers could use powerfull environmental modelling tools and existing, proven formulas to check that what they did is correct. For SA, it is straightly impossible because defining an architecture does not mean it will literally work - there is no prove! You probably model what it "might" do, or what performance it might get, but there is no guarentee that the whole thing will work as it should.
Just my 2cents...now that I read it again the things I said doesn't make much sense, lol. Anyway, good luck with your Blog!
I think you are right Hugh. In Civil Engineering you can measure everything and there are formulae for working out whether the specification is right or not. There are some ways you can do that with Software specifications as well, but I think there definitely has to be more work done in that area.
Vilas, is Alloy one of those tools for modelling SA? Or does it just model a problem?
The QWAN -> Very interesting and innovative, will have to keep up with developments there. It is very interesting that all patterns people quote Christopher Alexander. I should read his book.
http://www.qwan.com/
"Quality Without A Name"?
Our tools are still inadequate. There is nothing in Java that corresponds directly to a design pattern, for instance. It will be another twenty years (at least) before we can design programs the way we design houses.
Alloy -> I think it is more for formally validating a particular part of a system's logic. i.e it is at the implemenation level rather than the specification level.
Post a Comment