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

[FredK]

"Bill Gunshannon" <billg999 at cs.uofs.edu> wrote in message 
news:741fk5F11d01iU1 at mid.individual.net...
> In article <v9DhRs3z$Ghu at eisner.encompasserve.org>,
> koehler at eisner.nospam.encompasserve.org (Bob Koehler) writes:
>> In article <grdimo$dh0$00$1 at news.t-online.com>, Michael Kraemer 
>> <M.Kraemer at gsi.de> writes:
>>> Bob Koehler schrieb:
>>>
>>>>
>>>>
>>>>    UNIX is still a two-mode system which forks new processes every time
>>>>    it turns around, and has no concept of files beyond stream of bytes.
>>>
>>> And ? So what. Essentially it *is* a sack of bytes.
>>> Records grouped in blocks (as in MVS and VMS) are relics
>>> from the era of slow tape and disk drives which had to be accessed
>>> at a rather low level.
>>
>>    What's wrong with having the right tool for the job?
>
> What's right with making everyone carry around a 40 lb. toolbox when
> the only tool they really needed was a screwdriver?
>
>>                                                          Sometimes a
>>    stream of bytes is the right tool, sometimes it's not.  Generally
>>    adding data structures and organization helps solve a problem.
>
> But why make everyone put up with the additional overhead when one
> person needs more than a stream of bytes?
>
>>
>>    Which means having records and blocks, which are NOT in any way tied
>>    to the low level storage, is a good option, and an enhancement of the
>>    file system, not a relic.
>>
>>    Only having one way to do things is a relic of the late 1960's.
>
> Forcing people to buy and constraining them with all this overhead
> which they have no need for is even more a relic of the 60's.
>

I continue to see this criticism, and yet do not understand it - since the 
actual problem with VMS file IO isn't RMS per-se but the file system (ODS).

Yes, RMS is a large toolbox, but it's cost in general is minimal - mostly in 
file open semantics. Ultimately neither UNIX or VMS see disks as a "sacks of 
bytes" they see them as a collection of disk blocks - and RMS can read/write 
disk blocks just fine - and the semantics of accessing those blocks as 
"sacks of bytes" is trivial (ignoring cluster/stream locking semantics).

The real problem with VMS disk IO isn't RMS - you don't need to use RMS you 
could use QIO and layer anything you want as an access method.  The real 
problem is ODS which is an antique from a different era which can't compete 
with the performance of moden file systems, and cached file systems.