Tonight’s East-Side SeaFunc Meeting
Tonight we had a good turn out for the East-side version of the Seattle Functional Users Group (SeaFunc). The theme of the night was tools and IDE’s, but of course discussion was lively and covered many different topics. We had seven attendees in all, including Jeremy, a software developer from Intentional Software which was particularly appropriate and interesting.
We found out that Charles Simonyi, president/founder of Intentional Software recently published this paper on Intentional Software which appeared in OOPSLA 06. Frank Krueger was also kind enough to bring some printed out notes about common problems he faced with software development. What was interesting was how the responsibility for the issues he mentioned often straddled many areas of software development: language, tools, libraries, process, and programmer discipline. One issue Frank brought up that really caught my eye, and that clearly belonged in the domain of tool support, was the suggestion of unit-testing through example-centric programming. Frank and I also discussed the overlap of unit testing with Design by Contract. Design by Contract, Example-Centric Programming, and Extreme Programming are all bedfellows in my opinion.
I also asked at the meeting about how to best represent meta-data in comments in a source file. The topic of literate programming came up, and afterwards I did some research on documentation generation tools. This led me to believe that I probably want to add Doxygen or JavaDoc style meta-comments to the Cat language. Thanks to Frank for steering me away from XML.
There once was a programming language called PLUNQ that contained a markup language called UNQ as a subset. The project seems to have been discontinued. UNQ markup is very minimal and I always thought that it would be a good fit for doc comments.
Doxygen and Javadoc only support shallow structure. You generally need to taint the doc comments with HTML or XML for deeper structure. Out of curiousity, why did this Frank recommend against using XML? Was it because of its verbosity? If that was the case, UNQ or something similar to it could provide a valid alternative.
You can read about (PL)UNQ here:
http://web.archive.org/web/20050910011000/http://plunq.sourceforge.net/
Comment by Mikael Lind — June 21, 2007 @ 8:21 am
That is a very good point Mikael. The nested structure of useful meta-data brings us back to XML or something similar. Frank and I both thought that XML was too verbose, but this is due mainly to poor tool support. My other beef with XML is that it is overkill for lightweight marked-up textual data, and any usage would end up being non-compliant XML. If you are going to break the XML spec, might as well look at other specs that fits our needs better.
UNQ is interesting. I found another current application of it here: http://phantasmal.sourceforge.net/Operation/Phrases.html
I wonder if JSON for comment meta-data would be a good choice?
Another possibily is Wiki-markup, but I think that is probably better suited for mostly flat data with lots of symbolic links.
Comment by cdiggins — June 21, 2007 @ 3:17 pm