Re: Passing sockets with "givedescriptor()" (across user profiles)

On 3/5/21 10:36 AM, Scott Klement wrote:

So, indeed, not only was this an option 20 years ago, but...  it is his *requirement* that it works like a program from 20 years ago!

Not quite.

It
*is*
a program from 20 years ago. While there's no realistic expectation that anybody would be using it on anything older than V6R1, much less a requirement to actually *work* on anything older than V6R1, there *is* a realistic expectation that it *continue to work*.

Which it does. (And if it ain't broke . . .)

And we have better things to do with our time than spend a month or more re-engineering and re-validating the client sign-on protocol (which might affect not only the server side, but the lowest level of two separate clients) from scratch. I already had to do that once, when we went from SNA protocols (for which the OS, as I recall, automatically launched not one, but *three* server-side jobs per client sign-on) to TCP/IP (and from a Smalltalk client to a Java client). We'd much rather spend our time improving the user experience of *all* of our products.

--
JHHL
(Looking forward to spending yet another weekend with a Linotype machine that's around eight years older than I am -- did you know that Ottmar Mergenthaler was using binary codes to sort things back in the day when a "digital computer" was a guy counting on his fingers?)