Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!decvax!ucbvax!AC.UK!SYSMGR%UK.AC.KCL.PH.IPG
From: SYSMGR%UK.AC.KCL.PH.IPG@AC.UK.UUCP
Newsgroups: mod.computers.vax
Subject: FORTRAN segmented record type
Message-ID: <8701151932.AA24237@ucbvax.Berkeley.EDU>
Date: Thu, 15-Jan-87 14:32:40 EST
Article-I.D.: ucbvax.8701151932.AA24237
Posted: Thu Jan 15 14:32:40 1987
Date-Received: Fri, 16-Jan-87 00:00:09 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 25
Approved: info-vax@sri-kl.arpa

>     How does one determine from inside a program whether an existing disk file
> (FORM="UNFORMATTED")  is  RECORDTYPE="VARIABLE"  or  RECORDTYPE="SEGMENTED"
> (say from within a FORTRAN program).
>     My primitive solution to the problem has been to OPEN the file
> with  RECORDTYPE=SEGMENTED  and then do a test read with IOSTAT=IOS.
> If IOS=35 (segmented record format error) I CLOSE the file and re'OPEN it,
> this time with  RECORDTYPE=VARIABLE.  Otherwise I just REWIND the file and
> go on about my business.

It isn't possible to tell the difference between a file written by FORTRAN with
segmented records, and one with  ordinary VARIABLE records, with 100%
reliability. Reason is that SEGMENTED records are not generated by RMS
directly; they are specific to FORTRAN and are just ordinary RMS VARIABLE
records with extra data bytes thrown in to allow FORTRAN I/O to
reassemble the segments. To be 99.9% sure, set up a loop to do dummy reads
as above all the way to the EOF, but you still MIGHT have user data in a
VARIABLE file which happens to look exactly like SEGMENTED record control data.

If DIGITAL are listening --- this is rather silly. Even IBM support segmented
records! Why not include these as another sort of RMS record structure?

Nigel Arnot (Dept. Physics, Kings college, Univ. of London;  U.K)

Bitnet/NetNorth/Earn:   sysmgr@ipg.ph.kcl.ac.uk (or) sysmgr%kcl.ph.vaxa@ac.uk
       Arpa         :   sysmgr%ipg.ph.kcl.ac.uk@ucl-cs.arpa