Comments for Paul Findlay http://findlay.net.nz/paul Paul Findlay and his online content Wed, 07 Jan 2009 05:01:12 +0000 http://wordpress.org/?v=2.6.3 Comment on XML: Parser by Cowtowncoder http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-35 Cowtowncoder Fri, 12 Oct 2007 23:41:16 +0000 http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-35 Hey Paul, good luck with the development! Like I said, it's a non-trivial task. But I think using a state machine is a good idea (in case it wasn't obvious from my first comment), and it was something I was thinking about too. In the end choice had more to do with possibility to recycle some existing pieces, than feasibility of approaches. I hope to learn more about the implementation if and when you get that far. Hey Paul, good luck with the development! Like I said, it’s a non-trivial task. But I think using a state machine is a good idea (in case it wasn’t obvious from my first comment), and it was something I was thinking about too. In the end choice had more to do with possibility to recycle some existing pieces, than feasibility of approaches.
I hope to learn more about the implementation if and when you get that far.

]]> Comment on Fingers crossed by Dave http://findlay.net.nz/paul/2007/10/03/fingers-crossed#comment-34 Dave Wed, 03 Oct 2007 21:10:01 +0000 http://findlay.net.nz/paul/2007/10/03/fingers-crossed#comment-34 While you are finding something to do, you could work through all those books you have been amassing. While you are finding something to do, you could work through all those books you have been amassing.

]]> Comment on XML: Parser by Paul Findlay » XML: Impossible to be original http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-33 Paul Findlay » XML: Impossible to be original Thu, 19 Jul 2007 22:46:59 +0000 http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-33 [...] combination of these XPath expressions get applied like a giant state-machine/trie to the incoming XML events in something approximating log(n) time, so large numbers of XML streams can be handled [...] [...] combination of these XPath expressions get applied like a giant state-machine/trie to the incoming XML events in something approximating log(n) time, so large numbers of XML streams can be handled [...]

]]> Comment on XML: Parser by Paul http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-32 Paul Thu, 19 Jul 2007 14:05:40 +0000 http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-32 I'm glad you have done it, and thanks for the encouragement :) Woodstox is the parser mine looks up to.. I hope the FSM thing works out for me, since its somewhere around 400 states already (I could reduce that by making it a pushdown automata..), utf characters get decoded to Unicode only as they are needed. I plan on using what I call a "Tim Bray technique" where chars > 128 get transformed to corresponding ASCII characters for the sake of keeping the transition characters in the FSM low, but inside the state the original Unicode character is checked for its validity to be present in whatever part of the document we are looking at. Perhaps if profiling shows to much work is being done, I could make it happen over longer runs of characters. Yes, I want it to be a niche thing (and it should be just because it is written for the <a href="http://www.digitalmars.com/d/" rel="nofollow">D programming language</a>), and you did read my intentions right. I want the parser to be suitable for environments like jabber servers or handling thousands of concurrent REST requests etc. I'm not sure anyone cares to do this at such a low level, so I may end up building some recursive descent parser generator that fits with an asynchronous IO framework.. Hopefully someone wants to pay me somehow for the end usage to compensate for being the only person in the world who cares :) I’m glad you have done it, and thanks for the encouragement :) Woodstox is the parser mine looks up to..

I hope the FSM thing works out for me, since its somewhere around 400 states already (I could reduce that by making it a pushdown automata..), utf characters get decoded to Unicode only as they are needed. I plan on using what I call a “Tim Bray technique” where chars > 128 get transformed to corresponding ASCII characters for the sake of keeping the transition characters in the FSM low, but inside the state the original Unicode character is checked for its validity to be present in whatever part of the document we are looking at. Perhaps if profiling shows to much work is being done, I could make it happen over longer runs of characters.

Yes, I want it to be a niche thing (and it should be just because it is written for the D programming language), and you did read my intentions right. I want the parser to be suitable for environments like jabber servers or handling thousands of concurrent REST requests etc. I’m not sure anyone cares to do this at such a low level, so I may end up building some recursive descent parser generator that fits with an asynchronous IO framework..

Hopefully someone wants to pay me somehow for the end usage to compensate for being the only person in the world who cares :)

]]> Comment on XML: Parser by Cowtowncoder http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-31 Cowtowncoder Wed, 18 Jul 2007 20:29:50 +0000 http://findlay.net.nz/paul/2007/07/04/xml-parser#comment-31 Yes. But definitely doable. Now, assuming I understand you right, one main benefit would be the ability to do non-blocking parsing: that is, either get next event, or indication of "not enough data yet to know". For what it's worth, I actually have written core of such a parser, and one that does implement Stax API too. At this point the main challenge is figuring out who would care enough to want to get it ready for real use. ;-) There are already enough good-but-blocking parsers for this to be bit of a niche thing I suppose. Oh, also, I didn't use FSM, but that's mostly since it gets rather hairy when one wants to also integrate utf-decoding with parsing, for further performance gains. Yes. But definitely doable.
Now, assuming I understand you right, one main benefit would be the ability to do non-blocking parsing: that is, either get next event, or indication of “not enough data yet to know”. For what it’s worth, I actually have written core of such a parser, and one that does implement Stax API too. At this point the main challenge is figuring out who would care enough to want to get it ready for real use. ;-)

There are already enough good-but-blocking parsers for this to be bit of a niche thing I suppose.
Oh, also, I didn’t use FSM, but that’s mostly since it gets rather hairy when one wants to also integrate utf-decoding with parsing, for further performance gains.

]]> Comment on Hilarious by Paul http://findlay.net.nz/paul/2006/06/08/hilarious#comment-16 Paul Wed, 12 Jul 2006 14:15:31 +0000 http://findlay.net.nz/paul/2006/06/08/hilarious#comment-16 Pre Increment/Decrement (++i/--i) commit the action, then return its new value, whereas Post Increment/Decrement (i++/i--) return the current value, then commit the change. Pre Increment/Decrement (+i/-i) commit the action, then return its new value, whereas Post Increment/Decrement (i+/i-) return the current value, then commit the change.

]]> Comment on XML Apis by Dave http://findlay.net.nz/paul/2006/06/15/xml-apis#comment-8 Dave Thu, 15 Jun 2006 23:36:19 +0000 http://findlay.net.nz/paul/2006/06/15/xml-apis#comment-8 Paul, this is...an interesting post. Paul, this is…an interesting post.

]]> Comment on Excerpt from The Brothers Karamazov by Dave http://findlay.net.nz/paul/2006/03/14/excerpt-from-the-brothers-karamazov#comment-6 Dave Wed, 03 May 2006 03:41:59 +0000 http://findlay.net.nz/paul/?p=16#comment-6 I waded through Brothers a few summers ago and, although the going was tough, I'm glad I did it. Some of the large dialogues between the Ivan (I think) and Alexey made for compelling reading. Hey - have you read "The Da Vinci Code" yet? I thought I might do so whan I get my hands on a copy. I waded through Brothers a few summers ago and, although the going was tough, I’m glad I did it. Some of the large dialogues between the Ivan (I think) and Alexey made for compelling reading.

Hey – have you read “The Da Vinci Code” yet? I thought I might do so whan I get my hands on a copy.

]]> Comment on Protected: Confessional by Marcus http://findlay.net.nz/paul/2005/11/26/confessional#comment-5 Marcus Sat, 26 Nov 2005 02:51:13 +0000 http://findlay.net.nz/paul/?p=13#comment-5 Protected Comments: Please enter your password to view comments.

This post is password protected. To view it please enter your password below:

]]> Comment on My website is boring by Paul http://findlay.net.nz/paul/2005/10/04/my-website-is-boring#comment-4 Paul Wed, 12 Oct 2005 08:29:21 +0000 http://findlay.net.nz/paul/?p=7#comment-4 I won’t force alcohol down you, and the drinking is not because of website driven despair.<br /> <br /> I tried grey lions but there wasn’t enough <em>contrast</em>. I won’t force alcohol down you, and the drinking is not because of website driven despair.

I tried grey lions but there wasn’t enough contrast.

]]>