Hi, Ben! Thanks for the input. I went to the Parsec and Attoparsec parser
links. Attoparsec was new to me. From the Parsec link:
Combinator parsers are written and used within the same programming language
as the rest of the program. The parsers are first-class citizens of the
language, unlike Happy parsers, which must be generated via a preprocessor.
End Quote
I may want to go with Parsec or cousin just because the approach seems
elegant. What I am fishing for is something you only learn from experience
or learn from talking to people who have experience. My impression is if you
want speed, Happy is the way to go. I will be making an investment in one or
the other paradigm so I need to understand the relative merits of each in
order to make a decision.
I would think Happy is a dark place whereas Parsec is a place of light which
becomes important as correctness that I can personally attest to is
concerned. What I am getting at is my impression is with Parsec the concepts
involved are more important than the actual code itself. I suspect with
Happy you could understand the concepts involved and the tool will continue
to be necessary. I value the educational experience. Parsec or cousin may
provide a better educational experience. Is Parsec slow like a snail
compared to Happy? or are they similar? Parsec certainly seems more
flexible, but my question concerns performance. You know the embarrassing
facts that you won't find in the brochure.
Thank you this has helped clarify my thinking.
--------------------------------------------------
From: "Ben Lippmeier" <b...@ouroborus.net>
Sent: 10 Wednesday March 2010 1734
To: "John D. Earle" <johndea...@cox.net>
Cc: <haskell-c...@haskell.org>
Subject: Re: Fast Haskell Parser
Hi John,
Doing a Google search for "haskell parser" returns the following link as
its first result. That's the parser that GHC uses.
http://www.haskell.org/happy/
You could also check out the following:
http://www.haskell.org/haskellwiki/Parsec
http://hackage.haskell.org/package/attoparsec
This would also be a perfect question to ask on the haskell-cafe mailing
list...
Cheers,
Ben.
On 11/03/2010, at 10:39 AM, John D. Earle wrote:
I was thinking of ways to create an efficient Haskell parser. My initial
thinking was to write a top down parser in Haskell, but if you want speed
a table driven approach may make greater sense.
Due to the existence of build bots there is a certain degree of
compliancy concerning build times. I feel that convenience is an
important factor. It should be convenient to build the source. Build bots
make an assumption, namely the existence of a formal infrastructure. I
believe that it should be possible to build something from source
casually.
This is a less demanding goal than high performance incremental builds.
It would be nice to out perform make files because if you fail to do
this, can it really be said that you are making progress? Correctness
appears to be a low priority among computer programmers. That said, it
may be worth investing some time in advance to figuring out how to best
achieve both objectives, namely correctness and performance. Who knows
skills acquired in one project may be useful in another and performance
is usually welcome.
So my question is, What sort of tools and methodologies exist in Haskell
to create high performance parsers? My impression is the speed at which
the parser performs its task is not the bottle-neck, but the parser might
as well be designed to be efficient so as not to be intellectually lazy.
It may even turn out that the parser may need to be efficient merely to
compensate for the spawn of correctness, namely slow builds.
_______________________________________________
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
_______________________________________________
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
