Mass-Produced Software Components

leftpad.

It's a rich document to read in terms of thinking about those times and how it is like today. For instance when it comes to text parsing people very much use regex or something like lex or yacc and you can count on regex in the standard library.

There is the fact of having a system of distribution like the original CPAN

https://www.cpan.org/

which is independent of many other technologies such as namespaces and garbage collection that make that kind of thing scalable and practical. There is the Java kind of class-based 'object' but there's also the $-@-# 'holy trinity' of Perl of scalar, list and hashtable. 1960s languages had a 'symbol table' built into the compiler, today's languages put that data structure in the standard library and frequently build the interpreter around it.

In the form of JSON that data structure rules the world, but non-anemic 'objects' are optional in Componentware today, I mean I can write a package of components for React with or without class-based objects and the end user will not particularly care.

In my mind SYSGEN is the old 360's answer to the Spring Framework but less dynamic with a simple form of compile-time metaprogramming, like compiling the Linux Kernel with different options. Thing is, a modern OS for Linux or Windows or current zOS has dynamic device drivers and figures out the machine configuration at runtime, but it used to be that you baked in the configuration data and required modules at compile time.

Actually, (1969). The conference this is from was in 1968, but this wasn't published until the following year.