From:	MERC::"uunet!WKUVX1.BITNET!MacroMan" 17-APR-1993 14:42:59.74
To:	MACRO32@WKUVX1.BITNET
CC:	
Subj:	Re: Structure of relative files

In article <1993Apr16.180924.4401@cpu.com>,
    gwlester@cpu.com (Gerald W. Lester) writes:

> Can anyone give me a pointer to where the structure of a relative file is
> documented.  I need to be able to read some without the benifit of RMS.
>

   I'm not sure it's ever been "publically" documented. I have a copy of
an "RMS-11 Design and Logic Manual" from 1977 which documents it, but it's
not a regular DEC publication, it's something that looks like it was run
off on a line printer, with figures hand drawn (and typed ) in.

   Anyway, according to this manual there are 3 possible record formats
for a relative file:

     MRS = Maximum record size ( as stored in files record attributes )
     FIXED CTL = Size of Fixed Control Area ( for VFC records )

   Fixed length record ( cell = MRS + 1 )

       +-----+-------------------------------------------->
       | ctl |     Data
       +-----+-------------------------------------------->

   Variable Length record  ( cell = MRS + 3 )

       +-----+--------+----------------------------------->
       | ctl | size(2)|   Data
       +-----+--------+----------------------------------->

   Variable Length with VFC  ( cell = MRS + FIXED CTL + 3 )

       +-----+--------+-----------------+------------------>
       ! ctl | size(2)| HDR(1-255)      !   Data
       +-----+--------+-----------------+------------------>

   Records are stored in fixed length cells ( cell size as shown above )
   Note this applies even to "variable length" records, the data the user
   sees is of a variable length but the space used within the file is constant
   for each record.

   Cells are stored in buckets ( number of blocks in a bucket is defined
   by the BUCKETSIZE attribute for the file ). Cells do not cross bucket
   boundaries. There is no "overhead" space in the bucket ( ie the first
   data cell starts at the first byte of the bucket )

   In each case ctl is a single byte which indicates if the record in
   this position currently exists ( value 8 ), has been deleted
   ( value 4 ) or never used ( value 0 ).

   The first block of the file is the "prologue area". Data buckets begin
   at block 2. You will need one field from the prologue area, that is
   the "End of File Block", this is the last virtual block which has been
   initialized. You need this information so that you don't read past the
   logical end of file ( and into uninitialized blocks ). This value is
   a longword stored at an offset of 112 (decimal).

   I hope this information is helpful and accurate ( I've not worked with
   relative files much so I'm just quoting from the manual ). Let me know
   if you need any more information and I'll try to find it.

--
=============================================================================
Malcolm Dunnett    Malaspina College                Email: dunnett@mala.bc.ca
Computer Services  Nanaimo, B.C. CANADA V9R 5S5       Tel: (604)755-8738