Bob Sneidar wrote:

> I suppose thinking about it you could arrayencode an array then do a
> search on it

You probably don't want to do that. I've given a lot of thought to methods of indexing large LSON files on disk, and I can find no method anywhere near as practical as working with them in their native array structure, given their linear, non-indexed LSON format.

Besides, in order to serialize an LC array it needs to be small enough to fit into RAM to begin with, so the one thing we know about all LSON files is that they fit into RAM nicely. :)

It's been a while since Mark Waddingham generously provided some notes in the LSON format (for the older format, which has changed since the introduction of Unicode in v7), but IIRC it was roughly:

0x05 -- one-bye header indicating that what follows is an array (now 0x06 in v7 and later)
  <element type op-code, with 0x05/0x06 being an array>
     <element name> NULL <4-byte data length indicator><element data>

There is a different op-code for numbers than for strings (IIRC 0x02 for numbers), allowing numbers to use a more compact binary form.

I'm guessing that in addition to the new op-code indicating an array type, the former NULL separator between element name and length UINT4 has been replaced with a preceding length byte, since of course NULLs can be part of the Unicode string of the element name.

Nice tidy format, well suited for disk storage and network transfer, but looking for things requires linear search in LSON, whereas the de-serialized native array form takes advantage of the super-quick bucket hash to find a given key.

Much as XQuery works on an in-memory, already-parsed form of XML, searching associative arrays in memory will be the way to go.

So looping is both faster to execute and easier to script in array form.

The biggest challenges would be parsing the query expression, and generalizing evaluation within the loop(s) to handle the range of query options.


--
 Richard Gaskin
 Fourth World Systems
 Software Design and Development for the Desktop, Mobile, and the Web
 ____________________________________________________________________
 ambassa...@fourthworld.com                http://www.FourthWorld.com

_______________________________________________
use-livecode mailing list
use-livecode@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode

Reply via email to