home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!agate!doc.ic.ac.uk!warwick!uknet!yorkohm!minster!pete
- From: pete@minster.york.ac.uk
- Newsgroups: comp.programming
- Subject: Generating typical ``random'' programs
- Message-ID: <727969084.4760@minster.york.ac.uk>
- Date: 25 Jan 93 13:38:05 GMT
- Organization: Department of Computer Science, University of York, England
- Lines: 34
-
- The problem:
-
- I am working on a software tool which generates fault trees from
- programs written in an Ada subset. I want to test some ideas on how the
- size of the generated fault tree relates to properties of the input code
- fragments...
-
- Clearly the more programs I throw at it the better. I've put several
- examples through it and got interesting results, but what I really want to
- do is get hold of hundreds/thousands of legal programs of varying sizes, with
- differing mixes of statement types, to submit to the tool as input.
-
- We don't have that many programs, so I'll have to generate some which ``look
- like'' real programs from the point of view of the FTA tool. I think some
- kind of Markov-model approach based on analysis of existing programs
- might do the job -- it can generate reasonable travesties of human
- language -- but I'd be interested to know if anyone has ever done anything
- similar...
-
- Somewhere in the literature on test case generation there must have been
- work on the generation of legal but not necessarily meaningful programs which
- meet given criteria on densities of particular statement types and depth of
- nesting, but I haven't managed to find anything yet.
-
- Does anyone have any pointers to similar work?
-
- I'll summarise any responses to the Net.
-
- cheers,
- pete
- --
- *Peter Fenelon -- Research Associate -- Software Safety Assessment Procedures*
- Dept. of Computer Science, University of York, York, Y01 5DD (+44/0)904 433388
- EMAIL: pete@minster.york.ac.uk `There's no room for enigmas in built up areas'
-