Hi All,
I'm a longtime lurker, infrequent poster of this list.
I have inherited a project here which is porting a proprietary ISAM
library and BASIC runtime to Linux for a company to move off SCO (yay).
This phase has been completed, and I needed to learn a lot more about
ISAM than I wanted. As part of the project, I wrote a perl module in C
to access the ISAM data from outside of the proprietary system. From the
beginning, I had wondered if it was possible to provide an SQL view of
the data.
I am now looking at doing just that.
There are two ways that I can think of to go about it. I can either
modify the ISAM code in the system to interface with the SQL server,
sending updates, deleting rows, etc as the application calls the various
ISAM routines. Here's where my limited experience with database systems
shows. I am not sure how to implement things like scan forward, scan
backward etc. I am thinking of perhaps a big query that returns all rows
sorted by the key field, and then the actual scan commands to operate on
a type of cursor. Does this make sense?
The second way would be to make a SQL server understand how to read the
ISAM tables. Mysql has support for custom data stores, and looking at
the API it looks like I could probably implement a read only, no indexes
SQL view of the data in a day or so. This method ropes us into mysql.
Has anyone tried to 'emulate' ISAM access methods using SQL? Is it
feasable? I would prefer to stay away from option #2, but at this point
it certainly seems the easiest...
Thanks,
Tim