[Info-vax] BASIC compiler in the hobbyist distribution

[seasoned_geek]

On Tuesday, May 26, 2015 at 1:46:58 PM UTC-5, Stephen Hoffman wrote:
> 
> Windows and related packages are getting rather more secure, so the 
> attacks are working on the stuff that hasn't been locked down; the 
> add-ons.

They've reduced the size of the holes but they seem to have increased exponentially in number so instead of classic Swiss cheese it is now more modern, but insecure none-the-less.

> 
> There is good open source, and there is utter crap closed-source.

Windows would be a shining example of the utter crap closed-source.

> 
> The distinctions among compilers and interpreters and JITs really 
> aren't as interesting (at least to me, and I'd suspect most other 
> folks) as whether I can better incrementally test and see what's going 
> on with the code.   Near-continuous compilation and playgrounds -- gonzo 
> debuggers, with very simple interfaces -- and a decent IDE can make even 
> a compiler much more interactive, for instance.

IDEs can make some things nicer. I use QtCreator when working on embedded stuff and it is fine. That said, it doesn't have its own little universe. If you want to 'test out' a routine you have to either hack it into your current project and navigate through to test it or launch a shiny new project from scratch which is much more involved than another login session where one types BAS or some other single line command for a shiny new universe.

> 
> A more recent analog to that is an RPC, and there are various options 
> available.  OpenVMS even has one or two of these available.
> 
> If you're depending on invoking some code by line number as a form of 
> security and given the "attacker" is already at the command line and 
> able to test commands, well, that's not a very robust defense.  Now if 
> those line numbers were image-activation-specific cryptographic random 
> numbers and the address space was randomly shuffled around, that might 
> be a bit better.

It wasn't security from an attacker. It was security from an idiot. This was on PDPs and most didn't have modems. It was to ensure when the client stupidly tried to run programs from the command line instead of via the menu they would do no damage. We couldn't have big informative names like GENERATE_PACKING_SLIPS.EXE. We had to ship stuff like INV765 which was a totally different program from INV756. We also had a 64K-Word limit so when clients wanted modifications to INV750 and INV760 the naming standard had us walking forward 751, 752, etc.

> 
> Rexx?   Can't say that reaching back to the 1970s and to IBM mainframes 
> was something I'd particularly considered.  No UTF-8 support, and 
> adding that can be involved.   But if you do want Rexx, NetRexx 
> apparently compiles and runs on the JVM.

Ummm I haven't followed the ANSI standard for REXX, but the OpenSource Regina project added UTF-8 quite some time ago.

http://sourceforge.net/projects/regina-rexx/files/regina-rexx/3.8/


> 
> As for the planned x86-64 release of OpenVMS, how is your port off of 
> OpenVMS going, seasoned_geek?  Gonna keep those Itanium and older boxes 
> going for a while, while porting the applications off of the OpenVMS 
> platform?

I have no reason to leave. I'm still running Alpha because there never was a worthwhile Itanium chip.

This may shock you, but just a handful of weeks ago I got an email about a long term project porting a customer from VAX to Alpha. They had no interest in worthless emulation or watching an Itanium seer sizzle itself in a cabinet.

I would not be shocked to find out one client of my former company was __still__ running their business on that Micro-PDP we sold them during my last days at LIOCS. They put it in their lab quality room with filtered perfect power. Those boxes were built to last.

> 
> As a back-office server, yes, OpenVMS can work for some folks and 
> particularly for folks with existing OpenVMS applications.    One 
> challenge for VSI involves trying to grow beyond that base.

One thing which really blew my mind and should not have given the total incompetence at HP is the fact they didn't seize the moment with Service Oriented Architecture and ACMS. With a cluster and ACMS servers and a persistent message queue system they had the ultimate distributed SOA architecture in their hands and they simply sat there all stupified.