Comments on: A syntactic dilemma (and an intro to Gazelle’s ambiguity resolution) http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/ parsing, performance, minimalism with C99 Mon, 06 Feb 2012 23:44:51 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Berke Durak http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/comment-page-1/#comment-1171 Berke Durak Thu, 29 Jan 2009 16:44:15 +0000 http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/#comment-1171 Hi, Thanks for calling me "very smart"! I have just improved Aurochs' memory consumption by a factor of about ten. Speed is mostly unchanged. (Nothing too exotic - a carefully selected hashtable for storing the sparse memo table). On Javascript, it now takes 25 MB to parse 96 kB file, which is much more acceptable. I agree that the performance of PEG parsing is lower than what can be obtained by other methods, but I find it quite OK as a general-purpose parsing tool, especially in view of its simplicity. Hi,

Thanks for calling me “very smart”!

I have just improved Aurochs’ memory consumption by a factor of about ten. Speed is mostly unchanged. (Nothing too exotic – a carefully selected hashtable for storing the sparse memo table).

On Javascript, it now takes 25 MB to parse
96 kB file, which is much more acceptable.

I agree that the performance of PEG parsing is lower than what can be obtained by other methods, but I find
it quite OK as a general-purpose parsing
tool, especially in view of its simplicity.

]]> By: orib http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/comment-page-1/#comment-1128 orib Wed, 05 Nov 2008 19:07:21 +0000 http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/#comment-1128 Are you aware of Ometa / Pymeta and friends? It's PEG based (which I understand you don't particularly like), but an in-browser JavaScrript implementation takes negligible memory longer than the input itself. And it makes extending and switching parser extremely simple. Even if you disagree with the whole PEG idea, reading Alex Warth will be worth your time. Are you aware of Ometa / Pymeta and friends? It’s PEG based (which I understand you don’t particularly like), but an in-browser JavaScrript implementation takes negligible memory longer than the input itself.

And it makes extending and switching parser extremely simple. Even if you disagree with the whole PEG idea, reading Alex Warth will be worth your time.

]]> By: Andrew http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/comment-page-1/#comment-1122 Andrew Fri, 03 Oct 2008 04:35:51 +0000 http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/#comment-1122 I'd like to second the // option. Given that an empty regex is not at all useful (at least, I can't think of any use for one), the syntax can remain unambiguous. Looking at the rules: s -> a // b; s -> a / / b; and s -> a /x / b; s -> a / x / b; it would be a good idea disallow whitespace in the // operator, just to make things even more clear. I’d like to second the // option. Given that an empty regex is not at all useful (at least, I can’t think of any use for one), the syntax can remain unambiguous. Looking at the rules:

s -> a // b;
s -> a / / b;

and

s -> a /x / b;
s -> a / x / b;

it would be a good idea disallow whitespace in the // operator, just to make things even more clear.

]]> By: Dennis http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/comment-page-1/#comment-1070 Dennis Sat, 23 Aug 2008 18:54:11 +0000 http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/#comment-1070 This might be a lame idea, but could you use // for the prioritized choice? I can not think of a reason why you would want to use // when defining a regular expression. Everybody knows that two is better then one. Also, | is OR sometimes and || is OR other times, so why not / sometimes and // other times.... This might be a lame idea, but could you use // for the prioritized choice? I can not think of a reason why you would want to use // when defining a regular expression. Everybody knows that two is better then one.

Also, | is OR sometimes and || is OR other times, so why not / sometimes and // other times….

]]> By: Anonymouse http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/comment-page-1/#comment-1050 Anonymouse Mon, 18 Aug 2008 17:11:57 +0000 http://blog.reverberate.org/2008/08/10/a-syntactic-dilemma-and-an-intro-to-gazelles-ambiguity-resolution/#comment-1050 re. prioritised choice: How about |> re. prioritised choice: How about |>

]]>