3 ms·

Ask HN: What features from old programming languages are worth studying today?

12 points by Decabytes 3y ago

PaulHoule 3y ago
See https://en.wikipedia.org/wiki/PL/I https://en.wikipedia.org/wiki/PL/I which was IBM's attempt to make a language that would take the place that C did.
- mitchbob 3y ago
  And https://m-cacm.acm.org/magazines/2023/11/277418-lessons-from-pl-i-a-most-ambitious-programming-language/fulltext https://m-cacm.acm.org/magazines/2023/11/277418-lessons-from...
jantypas2 3y ago
That depends on what we mean by feature of course.... let's assume we mean concepts... here's my list: Believe it or not, Cobol's formatting for printed reports -- we still do print and formatted output worked well there. Modula-2s imports -- C++ is getting there, but imports just worked. Lisp's macros had their day. I think we could benefit from stepping away from "make a better C++" and ask "what is this language trying to do?" There's no universal "best" language, but one immediate benefit I'd like to see -- let's finally agree on a universal calling convention, at least on Linux for example, so every language can call another? Let's give credit to Microsoft's CLR on that one. Also, golang's getting the idea with structure tags, but how about structs that serialize to a universal format -- see Corba is not dead!
- PaulHoule 3y ago
  I want a really nice macro assembler for people who love assembly language (not gas!) tied to a theorem prover. It runs into the problem though that you kinda need portable code in 2023.
runjake 3y ago
I'm no CS student, but getting familiar with Lisp was useful for a deeper understanding of a bunch of stuff. I resisted looking into Lisp for a long time because of the HN hype. I still don't use Lisp for any real world projects, but it was useful for helping me better understand concepts in other languages, particularly JavaScript and Python. Learning SmallTalk was also helpful with regard to message passing.
karmakaze 3y ago
I would focus on "lisp without the parentheses" languages like F# or OCaml. Even when I write in any other language, I think by composing and chaining operations without considering the runtime machine (the internal hows). Eliminating time from cognitive load is the greatest cheat. Being able to write programs like correct SQL result set first, then refine for performance would be nice. Most processes can be modelled as input, acyclic series of boxes, output. Put persistence at the ends and the language is largely an implementation detail, as is the choice of persistence.
JoeMayoBot 3y ago
Assembly language might be interesting. Here's a free e-book, Assembly Language Succinctly, that will give you a quick overview: https://www.syncfusion.com/succinctly-free-ebooks/assemblylanguage https://www.syncfusion.com/succinctly-free-ebooks/assemblyla...
mikewarot 3y ago
I think that the "color" aspect of ColorForth[2] could have wider use. I view it as a lightweight version of literate programming[1], in that you can intersperse code with the description of it. It makes it quite clear what is code, comment, constants, etc. I think that the ideas that were brought forth in MetaMine[3] where quite interesting. The ability to mix declarative and imperative code in the same program, without melting your brain doing so, was a very powerful tool. I'm sad that it hasn't resurfaced. (I did find a fork of it on GitHub though[4]) [1] https://en.wikipedia.org/wiki/Literate_programming https://en.wikipedia.org/wiki/Literate_programming [2] https://en.wikipedia.org/wiki/ColorForth https://en.wikipedia.org/wiki/ColorForth [3] http://web.archive.org/web/20201014024057/https://www.youtube.com/watch?v=RvRc3c9TQ6c&gl=US&hl=en http://web.archive.org/web/20201014024057/https://www.youtub... [4] https://github.com/caminadab/metamine https://github.com/caminadab/metamine
max_ 3y ago
I think the language that has really caught my attention is the APL family. J, K, BQN etc

12 points by Decabytes 3y ago

See https://en.wikipedia.org/wiki/PL/I https://en.wikipedia.org/wiki/PL/I which was IBM's attempt to make a language that would take the place that C did.

mitchbob 3y ago
And https://m-cacm.acm.org/magazines/2023/11/277418-lessons-from-pl-i-a-most-ambitious-programming-language/fulltext https://m-cacm.acm.org/magazines/2023/11/277418-lessons-from...

jantypas2 3y ago

That depends on what we mean by feature of course.... let's assume we mean concepts... here's my list: Believe it or not, Cobol's formatting for printed reports -- we still do print and formatted output worked well there. Modula-2s imports -- C++ is getting there, but imports just worked. Lisp's macros had their day. I think we could benefit from stepping away from "make a better C++" and ask "what is this language trying to do?" There's no universal "best" language, but one immediate benefit I'd like to see -- let's finally agree on a universal calling convention, at least on Linux for example, so every language can call another? Let's give credit to Microsoft's CLR on that one. Also, golang's getting the idea with structure tags, but how about structs that serialize to a universal format -- see Corba is not dead!

PaulHoule 3y ago
I want a really nice macro assembler for people who love assembly language (not gas!) tied to a theorem prover. It runs into the problem though that you kinda need portable code in 2023.

runjake 3y ago

I'm no CS student, but getting familiar with Lisp was useful for a deeper understanding of a bunch of stuff. I resisted looking into Lisp for a long time because of the HN hype. I still don't use Lisp for any real world projects, but it was useful for helping me better understand concepts in other languages, particularly JavaScript and Python. Learning SmallTalk was also helpful with regard to message passing.

karmakaze 3y ago

I would focus on "lisp without the parentheses" languages like F# or OCaml. Even when I write in any other language, I think by composing and chaining operations without considering the runtime machine (the internal hows). Eliminating time from cognitive load is the greatest cheat. Being able to write programs like correct SQL result set first, then refine for performance would be nice. Most processes can be modelled as input, acyclic series of boxes, output. Put persistence at the ends and the language is largely an implementation detail, as is the choice of persistence.

JoeMayoBot 3y ago

Assembly language might be interesting. Here's a free e-book, Assembly Language Succinctly, that will give you a quick overview: https://www.syncfusion.com/succinctly-free-ebooks/assemblylanguage https://www.syncfusion.com/succinctly-free-ebooks/assemblyla...

mikewarot 3y ago

I think that the "color" aspect of ColorForth[2] could have wider use. I view it as a lightweight version of literate programming[1], in that you can intersperse code with the description of it. It makes it quite clear what is code, comment, constants, etc. I think that the ideas that were brought forth in MetaMine[3] where quite interesting. The ability to mix declarative and imperative code in the same program, without melting your brain doing so, was a very powerful tool. I'm sad that it hasn't resurfaced. (I did find a fork of it on GitHub though[4]) [1] https://en.wikipedia.org/wiki/Literate_programming https://en.wikipedia.org/wiki/Literate_programming [2] https://en.wikipedia.org/wiki/ColorForth https://en.wikipedia.org/wiki/ColorForth [3] http://web.archive.org/web/20201014024057/https://www.youtube.com/watch?v=RvRc3c9TQ6c&gl=US&hl=en http://web.archive.org/web/20201014024057/https://www.youtub... [4] https://github.com/caminadab/metamine https://github.com/caminadab/metamine

max_ 3y ago

I think the language that has really caught my attention is the APL family. J, K, BQN etc