[Info-vax] HP wins Oracle Itanium case

[Stephen Hoffman]

On 2012-08-22 13:19:19 +0000, Bob Koehler said:

> In article <k10u46$iuh$1 at dont-email.me>, Stephen Hoffman 
> <seaohveh at hoffmanlabs.invalid> writes:
>> 
>> o RMS files don't defragment themselves, and indexed files don't pack
>> old records.  The programmer or the end-user has to deal with that mess.
> 
>    I've got lots of file systems that don't defrag themselves.

So we should accept old limits?  Or should I get off your lawn?  :-)


> 
>> o RMS files require an extra license and related coding for
>> transactional processing and rollbacks, and the transactional API
>> doesn't document any means of synchronization with the other activities
>> that might be necessary within an application.
> 
>    RMS is not a DBMS.

Of course RMS is not a DBMS.

And few would want to use RMS when a SQL or NoSQL database is better choice.

And these days, I find myself writing (more) code when RMS is involved.

Or hauling around a SQL or NoSQL database, when that's more 
appropriate.  Or force-fitting RMS, where that's not (as you've 
mentioned) a DBMS.

The distinction between RMS and a DBMS matters not to me save in their 
relative capabilities; they're just different ways to store data bytes 
somewhere durable.  Or preferably store and retrieve the objects, and 
not have to mess around with the object graph/records/whatever.

For many applications, RMS is too limited.  For the reasons cited.  For 
others, a Unix byte stream would work just fine.


-- 
Pure Personal Opinion | HoffmanLabs LLC