|Looking for sample scanners! email@example.com (jansenpg) (1996-08-28)|
|Date:||28 Aug 1996 20:52:01 -0400|
|Organization:||Philips Research Laboratories|
Dear Compiler Constructor,
Almost every scanner for a high-level language looks globally the same.
You have a definition for identifiers, strings, integers, floats, chars
and bools and possibly also for hexadecimals or octals. Moreover, comments
and whitespaces needed to be skipped. These also come in a small number of
flavours: single line comments (with only some kind of start symbol) and
multi-line comments (with a start and end symbol). Multi-line comments can be nested or
not, for instance.
ONLY A SMALL SUBSET OF THE POSSIBILITIES OFFERED BY SCANNER GENERATORS IS
For our compiler generator tool set Argos which is specifically aiming at
FAST, EASY and RELIABLE compiler construction, we want to substitute the
old-fashioned scanner specification by a nice form where you can click on
various options for identifiers, strings, comments and so on.
In order to know exactly what is usually specified in scanners, we need to
look at a lot of existing scanners. My question to you is
COULD YOU SUPPLY US SOME OF YOUR SCANNER SPECIFICATIONS?
We prefer scanner specifications that are specified in a scanner generator
format like lex, flex, pccts, gmd, eli, ... It is hard for us to extract a
proper scanner specification from a home-brew C program.
This is very useful for us!
Please also send us your ideas on whether this is possible, whether you already
did some investigation in this area or just let us know additional (wild) ideas
in this direction.
Many, many thanks in advance! We will let you know the results of our
investigations in this same newsgroup.
Philips Research Laboratories
Prof. Holstlaan 4
5656 AA Eindhoven
phone: +31 40 27 44 659
[I never thought that the scanner was a particularly hard part of the
compiler, so I wonder if this would just end up making something that's
already pretty easy now very easy. -John]
Return to the
Search the comp.compilers archives again.