8 ms·
I disagree strongly with the authors assessment of the state of software development a few decades ago. By the late 1960s, we had realised software development
by kqr 5y ago
I disagree strongly with the authors assessment of the state of software development a few decades ago.
By the late 1960s, we had realised software development was hard. We put some good brains on the problem. Discussions between leading experts brought up several very important points, that we struggle with to this day:
- Naming things,
- Low coupling and high cohesion,
- Communication between developers,
- Communication with customers,
- Evolving prototypes,
- Avoiding the planning fallacy,
- Estimation,
- Support and understanding from upper management,
- and much, much more.
These are problems we struggle with today, but it's also the problems identified by and worked on by the experts of the late '60s into the mid '70s. We can learn a lot from what they discovered. (I sure have -- and I keep learning more!)
In fact, if I'm being a bit uncharitable, only three things truly seem to have changed since the late '60s:
- We have faster computers.
- We have virtual computers.
- We have higher level languages.
- We have version control.
Other than those four things, all advances in how to build software I've seen the last few decades are, in some sense, a rehash of what they found out early on.
- urthor 5y agoHonestly there is a bit more than that. The idea of mass development with large, open ended access to code is actually a surprisingly recent idea. Plus of course Stackoverflow has significantly changed the approach. Concurrent programming is also significantly different to its 1960s iteration. COBAL and Fortran did not even attempt to do such a thing, the handling of all those asyncs was not around. Many many other architectural details have also been implemented that were specifically designed not to be exposed to give classically trained programmers the illusion that things under the hood have not changed. However, they have significantly. Branch prediction leaps out at me, but also many other forms of evaluation, memory and processor design. There is more than you describe that has changed. If you would say that compilers and parsers have not changed, that would be extremely accurate. But there's a bit more to the world on either side of the compiler.
- kqr 5y agoMaybe the nature of our concurrency problems has significantly changed, but from what little I remember, I don't think so. More people attempt to design for concurrency now, but shared memory, actors, barriers etc have been around at least since the early--mid '70s. Do you really think branch prediction has significantly reduced the difficulties in developing software, or altered the way software developers work? Bringing up things that have changed behind a compatibility layer that makes us program against an emulated PDP-11 sounds to me like support of my thesis, if anything: the only changes to the way we work are faster computers and higher level languages.
- int_19h 5y agoConcurrent is not quite the same as async, though. If we're talking about async - especially the most common modern variety which is basically coroutines with some syntactic sugar - then it first appeared in mid-1960s in Simula. Although I don't think it was used much then.
- b0afc375b5 5y ago> only three things truly seem to have changed since the late '60s: > - We have faster computers. > - We have virtual computers. > - We have higher level languages. > - We have version control. > Other than those four things I see off-by-one errors hasn't changed since the '60s.
- fulafel 5y agoVirtualization came around in 1972 so was not far away either!