Treating keywords as identifiers where possible

anton@mips.complang.tuwien.ac.at (Anton Ertl)
Wed, 19 Jan 2011 14:45:28 GMT

          From comp.compilers
Related articles
language design implications for variant records in a pascal-like lang noitalmost@cox.net (noitalmost) (2010-12-22)
Re: language design implications for variant records in a pascal-like gneuner2@comcast.net (George Neuner) (2010-12-23)
Re: language design implications for variant records in a pascal-like bobduff@shell01.TheWorld.com (Robert A Duff) (2010-12-31)
Re: language design implications for variant records in a pascal-like gah@ugcs.caltech.edu (glen herrmannsfeldt) (2011-01-02)
Re: language design implications for variant records in a pascal-like bobduff@shell01.TheWorld.com (Robert A Duff) (2011-01-06)
Re: language design implications for variant records in a pascal-like cfc@shell01.TheWorld.com (Chris F Clark) (2011-01-14)
Re: language design implications for variant records in a pascal-like 9cn6w6402@sneakemail.com (Peter Canning) (2011-01-18)
Treating keywords as identifiers where possible anton@mips.complang.tuwien.ac.at (2011-01-19)
Re: Treating keywords as identifiers where possible robin51@dodo.com.au (robin) (2011-01-22)
| List of all articles for this month |
From: anton@mips.complang.tuwien.ac.at (Anton Ertl)
Newsgroups: comp.compilers
Date: Wed, 19 Jan 2011 14:45:28 GMT
Organization: Institut fuer Computersprachen, Technische Universitaet Wien
References: 10-12-040 10-12-043 11-01-005 11-01-009 11-01-027 11-01-057 11-01-086
Keywords: parse, design, LL(1)
Posted-Date: 21 Jan 2011 22:06:17 EST

Peter Canning <9cn6w6402@sneakemail.com> writes:
>On 1/14/2011 2:04 PM, Chris F Clark wrote:
>> From a grammar writing point of view, it is not particularly
>> difficult to introduce PL/I style keywords to any LR grammar. The
>> same changes should generally work for LL grammars also. If people
>> are interested I can document them here.
>
>I'm definitely interested in seeing a description of how to do this (in
>my case for LL grammars).


It's pretty easy for LL(1) grammars: At each position, treat all
keywords as identifiers if they don't occur in the first set. E.g.,
in many languages "IF" occurs only as keyword at the start of a
statement, so in a PL/I-style variant of those languages "IF" could be
treated as identifier everywhere else.


A disadvantage of this approach is that some syntax errors can only be
discovered later. E.g, to pick up the example above, if the semicolon
in front of an IF is missing, a Pascal compiler will notice a syntax
error when it sees the "IF", but not necessarily a compiler for a
PL/I-style language. I don't know how relevant this is in practice.
If only new keywords (for language extensions beyond the original
standard, where new keywords often cause problems) are treated in this
way, this disadvantage is probably small compared to the benefit.


Such a scheme is probably best implemented in the parser (which knows
about the first-sets), but at the interface to the scanner (so the
higher levels don't need to have special treatment of such
identifiers).


- anton
--
M. Anton Ertl
anton@mips.complang.tuwien.ac.at
http://www.complang.tuwien.ac.at/anton/


Post a followup to this message

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