LexAGen

There has been some recent discussion about lexical analyser generators
in this newsgroup. You might be interested in trying our new scanner
generator which is called LexAGen. LexAGen generates C-code scanners
which are slightly slower than GLA scanners (96% as fast) and like GLA
uses direct execution instead of table interpretation.


LexAGen is an interactive scanner generation environment so that
syntactic and static semantic specification errors are reported
as you make them. In addition, LexAGen builds an internal automata
representation incrementally as you specify the scanner. You can
test the partially specified scanner at any time inside of the
environment and edit individual specification productions.


The only catch is that LexAGen is implemented in Smalltalk-80 so you will
need Smalltalk-80 to run it. (I assume that it would require extensive
modifications to run it on any other dialect of Smalltalk since it makes
extensive use of MVCs). Note that although you need to have Smalltalk-80
to run it, you don't need to know Smalltalk to use it.


I am willing to e-mail the Smalltalk-80 source code for LexAGen to anyone who
sends me a direct request by e-mail. I am also willing to send copies
of the users manual (a technical report) by post to people who want to
use LexAGen.


Since LexAGen is research product, it is not production quality so you
are warned ahead of time that there may be some bugs!


Duane Szafron
Programming Languages and Methodologies Group
Department of Computing Science
University of Alberta
Edmonton, AB CANADA T6G 2H1


email: duane@alberta.uucp
--

Return to the comp.compilers page.
Search the comp.compilers archives again.