[Info-vax] RMS, was: Re: Assembly languages

[Arne Vajhøj]

On 4/11/2022 11:01 AM, Jan-Erik Söderholm wrote:
> Den 2022-04-11 kl. 16:13, skrev Dave Froble:
>> On 4/11/2022 9:06 AM, Simon Clubley wrote:
>>> Also, RMS indexed files were a neat solution for the 1970s/1980s, but
>>> in todays's world they also have not stood the test of time.
>>>
>>> For example, you work on a punched card model as there's no field level
>>> data structures in the RMS indexed file metadata, and unlike SQL 
>>> databases,
>>> you can't extend an RMS indexed file to add a new field, or change the
>>> size of an existing field, while the applications are running.

>> Your statement is just wrong.
> 
> What part of it? I find it mostly correct. A standard database product
> (we can pick Rdb since this is VMS) is just so much easier to deal with
> then a bunch of RMS files. And that is straight out-of-the-box, with no
> home-made tools to completment with.

NoSQL databases of key value store type (which is what RMS indexed files
are in modern terminology) are still used for specialized purposes.

The standard/default approach is and should be a relational database,
but sometimes requirements are not to the standard solution.

Rdb is a pretty big database to replace RMS indexed files, but it
will work. I would consider SQLite to be a more obvious replacement.
But options are always good. And in the middle there is MySQL/MariaDB.

>> Anything that can be done in any database product can be done, and as 
>> RMS is a database product, such things could be implemented in RMS.  
>> The fact that they are not at this time does not indicate that they 
>> could not be done.
>>
>> I will comment that the RMS developers did far less than they could 
>> have done back in the day.  Before RMS existed the company I worked 
>> for at the time had a database product that while file oriented, had 
>> the data field metadata and utilities that used it.  It was doable 
>> back then, the RMS developers just failed to do it.
>>
>> Not a fan of RMS ...
> 
> As Simon indicated, RMS (indexed) files has passed their best-before date.
> I do not see anyone writing a new application based on RMS indexed files.

The concept still exist and are used in the industry.

The biggest problem with RMS indexed files is probably the RMS part
not the indexed part. Max 32K records, VMS IO etc..

I don't see any good reasons for having a new VMS application
use RMS indexed files. The mentioned specialized purposes are
not something that will use VMS.

But I am also sure that some new VMS application will use
RMS indexed files, simply because of tradition and skills.

Arne