---- Original Message ----
From: "Donna Garverick" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, November 26, 2002 12:13 AM
Subject: [HP3000-L] migrating and rio files?
> hi all!
>
> well, wouldn't you know....for the *one* mpe application
> that we're actually going to migrate (vs. throwing away
> <sigh>) to unix -- we've hit a very good question -- what do
> you do with rio files? got me (probably should have never
> used them to begin with...but that's my opinion). the
> description i got from one of the developers is -- they
> sorta use these files (i guess there's lot of them) like
> light-weight ksam files (don't ask why.....) my suggestion
> to incorporate this information into their new (oracle, btw)
> database got shot down (we'd have to redesign the whole
> application <sigh>).
>
> so...has anyone crossed this particular bridge? - d
I'm not quite sure what constraints you are under, Donna, but before I ever
used an HP3000, I used a machine with Relative files, and did what I now
recognise that RIO files do.
We made the first byte a Status flag, and turned it On when we wrote a new
record, Off when we (logically) deleted one. Turning a record 'Off' didn't
affect its relative record position, or that of the records after it. All the
processes we read the file with knew to skip the deleted records.
Skipping the deleted records is done for you with RIO records, if you FREAD
them with RIO access. Likewise the status flag is 'outside' the formal file
definition.
But you can do the rest of what we did - direct access, deletion, replacement,
with 'standard' MPE files and the MPE file intrinsics (and even COBOL if I
read the manual aright). And so, I'd guess, with the analogue of 'standard'
files in pretty much any other filing system too, on a different platform, and
in its supported languages also.
Add the explicit status flag field, add explicit skips of records with status
Deleted, change RIO Deletes to conventional updates of the status flag.... is
this doable for you?
Mind you, we didn't stop at what we now know as RIO. We had ISAM, much like
KSAM, but it wouldn't do quite what we wanted....
Then we found the machine also supported 'Direct' files - these were relative
files, but which could have a designated key field, and you could search on
that key value to get the relevant record, via an internal hashing mechanism
in the OS. It was neat - it had secondaries of course, but organised matters
so they didn't migrate.
By using these Direct files, plus designating a few pairs of fields in some
Relative files to hold record pointers (forward and backward) and starting the
actual use of the records from record 1, initialising the unused records in
the files to be chained off record 0, we were able to do explicitly what we
now recognise as the province of....... a very well-known HP3000 subsystem :-)
--
Roy Brown
Posting with the OEnemy, tamed by OE-QuoteFix 1.17.6
http://jump.to/oe-quotefix
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|
