[antlr-interest] A postmortem of my use of antler

[Stuart Watt]

There is a pretty strong psychology of programming community in various
countries, and maybe this is the kind of thing some of them might be
interested in looking at. Some parts will already have been studied - a
colleague of mine did a lot of work on misconceptions of backtracking in
Prolog, basically looking at control flow in rule-based languages. He did a
lot of recording (protocol analysis and coding, errors, etc.) and rebuilt a
Prolog debugger which was significantly better (in both the technical
statistical sense and the commonsense one). 

This is not especially ANTLR-specific, as *exactly* the same point applies
to similar tools, e.g., bison/flex, Prolog DCGs and so on. So the methods
exist and can be used; however, doing it right can be extremely intensive
(in the case of Prolog, gathering the data alone took two months, ensuring
that the problems were well-specified, and so on, i.e., that you are not
investigating the task but the tool. Transcription, coding, and analysis
took considerably longer. 


This is a great suggestion, and an interesting possible project. I am sure
some funding agencies might love it, and it would be possible to put a good
case - part of which is the size of the community of ANTLR users. Actually
doing it requires people who can both understand development and have strong
research methods skills, and I have met relatively few of these, more's the
pity because they are immensely valuable. 

All the best
Stuart

-----Original Message-----
From: Benjamin Shropshire [mailto:shro8822 at vandals.uidaho.edu]
Sent: Monday, March 10, 2008 3:29 PM
To: Antlr Interest
Subject: [antlr-interest] A postmortem of my use of antler

...
What wold do the most good would be to have about 
1 or 2 dozen new users (a compilers class?) be thrown at antlr with 
nothing but the current docs to work from. let them play with it for 
about a day or two and then start asking them what they found 
interesting/useful/confusing. Then start answering there questions and 
/recording/ what they asked, what they wanted to known (when these are 
different that is valuable info). Also take notes on what kind of 
assumptions, correct and incorrect they made. This info would be of huge 
value to the ANTLR project.
...