[Info-vax] OpenVMS Development Annoyances

[Stephen Hoffman]

On 2019-04-24 20:14:26 +0000, Dave Froble said:

> On 4/24/2019 1:16 PM, Stephen Hoffman wrote:
>> I'd rather have a bigger OpenVMS installer than one as primitive as the 
>> current one.
>> 
>> "Does this product have an authorization key registered and loaded?"
>> 
>> Um, you're a computer. Look it up. Then tell me if I need to deal with it.
> 
> Yes, I like that attitude.  However, when I tell the compiler to find 
> the variables, why does everyone claim that I should declare all 
> variables?
> 
> Hypocrisy I say!  Hypocrisy !!!!

Short answer:

LMF is inherently and necessarily late-binding, and the answers can and 
variously do change over time.

You're arguing against static binding, and the answers can't change over time.

Well, not short of monkey-patching.



Long answer:

Development and installation tasks are typically different roles with 
different expectations and different user interfaces and different 
folks involved.

Server installations are headed toward full automation.

Fewer prompts and more defaults is a benefit.

I find the OpenVMS installs utterly archaic; far closer to an RSX-11M 
SYSGEN than they should be.

Prompts are cases where the developers didn't or couldn't provide a 
default, or didn't or couldn't eliminate the question.  Or didn't think 
the could.

LMF is a marvelous example of inward-facing UI designs. Incredible 
flexibility and fully self-hosted, but at the expense of the end-user.

What's shown as the proper PAK in the usual LMF prompt for those 
installation tasks is quote often incomplete, as any of various 
different license PAKs can enable various products.

The number of permutations is a whole lot smaller with LMF and PAK 
checks, than exists with the guess-my-intent compilation during 
software development.

Xcode is pretty good at the guess-my-intent and the fix-me stuff is 
surprisingly accurate, and far beyond anything I've used on OpenVMS.

The PAKs that can authorize a product can and sometimes do change over 
time, which means the developer can't know which PAKs will enable a 
product.

I've chased around bad variable declarations for many years and in 
various apps, and have found more than a few latent bugs by cranking up 
the compiler diagnostics.

I've chased around more than a few bad user interface designs too, and 
with a number of the messes here being of my own designs.

 End-users can and often do think quite differently from developers.



-- 
Pure Personal Opinion | HoffmanLabs LLC