[Info-vax] Anyone interested in another public access system

[glen herrmannsfeldt]

Bob Koehler <koehler at eisner.nospam.encompasserve.org> wrote:
> In article <grql8e$26k$1 at naig.caltech.edu>, glen herrmannsfeldt <gah at ugcs.caltech.edu> writes:
 
>> It is for VMS, where the underlying hardware consists of
>> 512 byte blocks.  It is much more interesting for 
>> IBMs MVS, where the hardware allows variable block length
>> chosen by the program.  (Between 1 byte and the track length.)
 
>   On both VMS and MVS, I can setup the file system to use just about
>   any logical block size that I want.  On those, and on UNIX systems, the
>   underlying disk sector size is often 512 bytes and physical block sizes
>   multiple of those.  And when I set up UFS or something similar on UNIX, 
>   I can tell the UNIX kernel to use lots of different logical block and
>   fragment sizes.

On MVS the blocksize is rarely 512.  On current systems it
might be internally to the drive, but as visible to the I/O
channel it is user defined between 1 and tracksize.

CMS and VSAM use a file system based on fixed size blocks
(often 4096), but MVS BSAM, BPAM, and QSAM are still often
used.  This is very visible with the C ftell/fseek functions.
 
>   It's only the lack of organisation at the API level that forces 
>   UNIX programmers to think only in stream of bytes.

-- glen