Re: SGML to HTML with jade?

Subject: Re: SGML to HTML with jade?
From: "James Clark" <jjc@xxxxxxxxxx>
Date: Thu, 26 Jun 1997 11:59:48 +0700
>  Mark> Now I've got your attention, how about this index creation
>  Mark> nightmare? I guess we're all waiting for James to fill in some
>  Mark> of the gaps in Jade.
> 
> Surely that's the thing that really should be done externally to get
> it right 

Why?

I had a glance a Mark's code, and I could see two things that would make it
very slow.  Firstly it's using a O(n^2) sorting algorithm (insertion sort).
 Secondly it's mapping characters to integers once every time strings are
compared.  I would suggest instead:

- building a list of entries: each entry is a pair of a sort-key and a
sosofo; the sort-key is a list of integers (or maybe a list of list of
integers).

- sorting the list of entries using a merge sort.

The only major inefficiency with this is the mapping of characters to
integers, but this would be done exactly once for each index entry, so I
doubt it would be a big problem.   This could be avoided if Jade did a
better job of optimizing case expressions, or if I added a (unicode-code
char) external procedure that returned the Unicode code for the character.

James


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist


Current Thread