[Info-vax] Free Pascal for VMS ?

[DaveFroble]

seasoned_geek wrote:
> On Wednesday, May 9, 2018 at 10:36:40 AM UTC-5, Stephen Hoffman wrote:
>>>>> Software development on VMS definitely need to move to OO.
> 
> No. 
> 
> I've written a lot of C++ and done a ton of cross platform stuff. Wrote my first book on it back in the late 1980s/early 1990s.
> 
> https://books.google.com/books?id=cdx_nLaqMn0C&printsec=frontcover&dq=%22Zinc+it!%22&hl=en&sa=X&ved=0ahUKEwi5w4jno_zaAhUO24MKHQX8DFMQ6AEILTAB#v=onepage&q=%22Zinc%20it!%22&f=false
> 
> VMS does not need to go OOP.

Don't have an opinion about whether such capability should be provided, but I 
agree, VMS does not NEED OOP.


> What is the ONE target market for OpenVMS?

I know, I know, pick me ...

Provide an environment to run VAX/DEC/Compaq/HP/VSI Basic.

That's my choice, don't like it, show me where I'm wrong.

> OpenVMS __was__ "The OS With Up-times Measured in Decades."

That is far from the only thing VMS has.

> Is it just the people who are still running it?

To be determined ....

> If that is the case you need a TCP/IP software appliance which is a fleshed
out version of the TCP/IP software appliance you already have. It needs to have
outbound definitions and a greater ability to specify transport layer security
and the ability to define packet boundaries from begin/end tag to size to simple
newline. All existing languages can call an TCPIP_INIT() procedure to set up
some common data communications area and ASTs to be triggered when the appliance
has queued up a full packet. You also need to purge the Unix tcp/ip libraries.

Your term is "TCP/IP software appliance".  Ok, but whatever it's called, VMS 
needs to be able to communicate with others, and an easy to use, secure, method 
sure is needed.

> You need VAX BASIC pretty much the way it is, with the interpreter brought
back for testing routines which can be written and tested without a massive
environment.

Don't know why an interpreter is required.  Today, compiles and links are rather 
fast.  Even faster when one sets up a command procedure.

But yeah, lots of Basic users.  It's a rather good product.  Anything can be 
improved, and that should be looked at.

> You need ACMS, C++11 (maybe 14) with the same "level" C compiler, COBOL-85 and FORTRAN 90. Oh wait! You've already got COBOL-85 and FORTRAN-90 so you just need the C/C++. You also need to port ADA as it is because, OpenVMS has one of the few functional ADA compilers and there is a ton of code out there. Dupont just inherited a bunch of it from Dow.
> 
> You need a production quality relational database bundled with the base package. Gee, isn't that why DEC bundled the RDB run-time back in the day? Only this needs to be a full package. PostgreSQL is the only real choice. SQLite is nice for embedded systems, but, it has next to no data type enforcement. 
> 
> Now, if someone wants to endure the pain of creating an SQLite or MySQL front end for RMS indexed files which have multi-typed records and bitmapped fields, have at it...<Grin>

RMS files should be considered a declining feature.  Today a database is so much 
better to use.  And with a database, the definitions are included, don't need CDD.

> Each of the previously mentioned languages needs direct support for the bundled relational database. The simplest and most widely accepted "solution" here would be to come up with a version of SQLMOD which compiles for a PostgreSQL back end. Exiting applications could easily be ported, jettisoning Oracle RDB license and support contract, which could be wonderful and justify the cost of migration.

I'll leave this to the experts for the details, but yes, make database access 
easy to use.

VMS needs a database.  If PostGRE is the answer, then add numeric ranges to DLM 
and perhaps it can be implemented on VMS.

Anyway, nice rant, but, I think many already knew most of this.

-- 
David Froble                       Tel: 724-529-0450
Dave Froble Enterprises, Inc.      E-Mail: davef at tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA  15486