[Info-vax] "bad select 38" (OpenSSL on VMS)

[Stephen Hoffman]

On 2016-09-17 17:54:55 +0000, John Reagan said:

> On Friday, September 16, 2016 at 11:11:48 AM UTC-4, Stephen Hoffman wrote:
> 
>> 
>> And (in all seriousness) does anybody still need __VAXC?
> 
> __VAXC is what is set with /STANDARD=VAXC.  That doesn't mean "I'm a 
> VAX compiler".  That is __VAX.

I'm well aware.

I find the use of VAX C an excellent flag to identify C source code 
that tends to be unmaintained or ill-maintained, and of source code 
that's often buggy, unstable and/or contains security vulnerabilities.

> That said, I'm all for tossing /STANDARD=VAXC over the side, but such 
> things tend to hang on for a long time until you are forced.

Then go do it.  Have the courage of your convictions.

Establish and then communicate your future plans.   That and the 
ensuing discussions are part of what boot camp is useful for to VSI and 
to customers, after all.

In this case, announce the end of VAX C.   That VAX C code won't make 
it across to x86-64.

(I'm here assuming there'll be little new C work on OpenVMS I64 or earlier.)

If folks really need VAX C and if VSI has the cycles, make'm pay for it.

I'd likely also announce that specific identified existing (and 
deprecated) APIs are also subject to removal.

Then start providing replacements for other problematic APIs, and then 
deprecate those problematic APIs, and then remove them.

I'd be seriously tempted to announce the deprecation and eventual 
removal of DECnet, for that matter.

This'll be painful and it'll be a shock to some existing sites, but so 
is making meaningful advancements within the constraints of ~forty 
years' accretion of immutable errors and security vulnerabilities.

We all know the only outcome of the current historic-preservation 
strategy, after all.





-- 
Pure Personal Opinion | HoffmanLabs LLC