Re: [open-cobol-list] Structure for cobol indexed files

[eurlix]

Hi Fernando,

Le Sat, 28 Jun 2003 14:18:06 GMT
<address@hidden> a écrit:

> Hi, my name is Fernando Benedictti, i'm from Uruguay.
> I want to write a tool that receiving a SQL sentence, it can 
> search into a structure of a indexed file a recover data.
> Can anybody tell me, how i can study or where obtain the 
> structure of indexed file in cobol?
> 
> I know that is a variant of B-tree.
> 
For what I know, the structures of indexed files are more 
or less "proprietary" (and more "more" than "less") . 
I think that the more used is C-ISAM from INFORM*X (now I*M)
wich use two files, one X.idx for the keys and one X.dat for
the data.
Some time ago, Acu was using an unique file for both keys and data.
M*F, after using C-ISAM has is own isam ...
I don't know others versions that should be exist, but BerkeleyDB
(used by Open-cobol) is well documented if it is what you want.

In C-Isam, the file  ".dat" is a sequential file, with just the
deleted records padded to low-value and without order (the .idx
know where is the searched record).

Regards,
-- 
Alain Lucari (Eurlix)