[I am taking the liberty of trimming the body and editing the subject, here, to fit the subject change. For people wanting the original context, http://jsoftware.com/pipermail/chat/2016-May/006994.html should have it, for at least a while...]
On Thu, May 26, 2016 at 11:31 AM, Mike Day <[email protected]> wrote: > I got the impression from the radio that they were still using Cobol, but > that wasn't said explicitly. The article includes this comment: "the US > treasury also needed to upgrade its systems, which it [a report] said was > using "assembly language code - a computer language initially used in the > 1950s and typically tied to the hardware for which it was developed." > Not sure I believe that. Doesn't assembly language still sit there in the > background? This actually might take a bit of describing. First, addressing the literal question: assembly language is fairly direct way of generating machine code. It's an alternative to fortran, cobol, c, etc. etc. But, also, each machine variation needs something different from assembly language. That's one of the appeals of "higher level languages" - they let you sweep machine dependent issues under the carpet (not always to a good effect, but that's getting into a different discussion). Still, compilers can go straight from their "high level" language to "machine language" without going through assembly language on the way there. Or, they might have an intermediate language which is distinct from assembly language (gcc, for example, has RTL as its intermediate form. This is a lisp-like representation of small sequences of machine instructions and these data structures go through a variety of simple transformations before emitting machine code). That said, given that this was a radio report, and that it was about the use of old hardware (which presumably still works rather well) the implication might also be that they were referring to a specific assembly language - like something for the IBM 370 or the like. Some people nowadays might laugh at some of the specs on those old mainframes, but high spec machines nowadays tend to have a lot of failure modes and inefficiencies which your typical sales representative would not be interested in dwelling on (and quite probably doesn't even know to ask about). Oddly enough, some of these issues (for example, "big i/o" issues like throughput, latency, or maybe some other "real time" subjects) also tend to matter for modern entertainment producers. And other, related issues (like reliability and so on) matter quite a bit for medical and industrial contexts. Anyways... the bottom line when dealing with machines is not "how new is it?" nor is it "what does the radio say about it?", it's "does it get the job done?". And that's not for me to say - at least not in the context of whatever the Treasury is doing... (But, ok, maybe I have made a few choice statements about what they are doing at times. And maybe some of those times I might even have been close to right?) Thanks, -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
