[Info-vax] x86 Update 4/22/19

[Stephen Hoffman]

On 2019-04-27 23:50:53 +0000, Phillip Helbig (undress to reply said:

> In article <qa2m07$1ve8$1 at gioia.aioe.org>, =?UTF-8?Q?Arne_Vajhøj? 
> <arne at vajhoej.dk> writes:
> 
>> But given that ECC RAM today is about 6 dollars per GB, then it is 
>> difficult to justify spending days/weeks on reducing memory consumption.
> 
> Of course this plays a role.  20 years ago, I paid 500 guilders for a  
> 4-GB disk.  Today, disks are 10 cents per GB or whatever.  25 years 
> ago, pricing workstations, we figured 100 marks per MB of RAM.  Times 
> have changed.

The other direction—forward—is far more interesting than where we've 
been, both around costs and around app designs.  More than a 
few—probably most—OpenVMS apps are still designed for twenty years ago. 
 Folks developing for OpenVMS are often preserving those same designs 
and design limitations, too.  OpenVMS tends to force some of those 
designs too, given the vintage of many of its available APIs.

> On the other hand, when people can include whole libraries just because 
>  they can, even if most or all of the stuff isn't needed, code can be 
> unnecessarily complicated.  One needs to find a balance.

Pragmatically, one needs better ways to deal with the inevitable.  Apps 
and systems *will* get more complex.  Dependencies *will* increase.  
Constituent libraries and apps *will* need patches.  Apps *will* be 
found vulnerable.  Patches and updates *will* need to be deployed 
faster.  Apps and patches *will* need better and faster and 
increasingly automated deployments.  Because even if "one needs to find 
a balance", there's a very clear trend where one and all are headed, 
and one and all may not be (are not at all?) prepared for that trend 
with the present state of OpenVMS and its features, and of the designs 
and assumptions typical of the VSI and ISV and end-user developers 
involved.



-- 
Pure Personal Opinion | HoffmanLabs LLC