Rev G2b Record Hashing (AREV Specific)
At 10 NOV 1997 07:38:05PM B. Sargent wrote:
Has anyone ever revealed the Rev G2b record hashing algorithm so that Rev files could be read by outside applications?
At 12 NOV 1997 05:13PM Tracy Graves wrote:
Well, you could import the data to linear hash through AREV, then modify it enough that it can be seen in OpenInsight, then use Data warehousing to conver the data to any data source you have an ODBC driver for.
Otherwise, I definitely don't know how to get at it.
Tracy
At 13 NOV 1997 02:16AM Chris Vaughan wrote:
If you are talking RTP52, this can be done outside of REV.
The ROSMEDIA MAP will give the DOS File Name for the internal REV Name.
It is not hard to work out the file frame from the record key (ie Hash Algorithm).
The file frame structure and and overflow navigation is self evident to anyone who has used DUMP.
We have a Class Library for VB developers who want to access REVG2B (and AREV) files. It supports all I/O and array handling. Unfortunately, it is not free.
At 13 NOV 1997 11:41AM Bob Sargent wrote:
Thanks, this is exactly what I'm looking for.
So how DO you work out the file frame from the record key?
How do I get your Class Library or VB developers that can access Rev G2b files?
At 14 NOV 1997 04:39AM Charles Schmidling wrote:
Hashing algorithms are many and varied and I have never figured out how Rev does it.
Just as an idea, you could do your own sequential search of your ROS files. Even build an index if you wanted to. Probably just keep the frame number you found it in for later access since that's all Rev does. Then it runs sequentially anyway.
Now, mind you, these are for ROS files. LNK files are another mystery.
Some hints.
- Finding the file and native DOS name and Modulo are easy to locate in ROSMEDIA.MAP. There will only be as many MOD frames, but any number of extensions.
- To link a frame and its extensions, each DOS file is marked by the first byte of the file 0-no ext, 1-look for extension.
- After the first byte, each record is arranged as a 5 digit length of the record (including key), the key, @FM delimiter, then the record itself. Followed by the next 5 digit length, key - and so on, and so on….
This would allow you to leap-frog through each frame and extension, looking for your record. Even though linear search is slow, each DOS file cannot exceed 64K in length so makes for fast reads. And you have a FAST machine anyway, don't you?
Just thoughts
Charles Schmidling
DATASCAN SYSTEMS, Inc.
cbms@juno.com
At 17 NOV 1997 12:40AM Chris Vaughan wrote:
If you send me an email (cv@hcsaust.com.au) I can send you the hash algorithm.
If you go to the web site 'www.hexagon.com.au' you will find all you need to know about HexConnect - the VB product that provides access to Pick, RevG, Arev, Universewisome and related DAtaBases.