He also avoided a lot of the more advanced features of Algol68, he thought it too complex, when he designed Pascal
He also avoided a lot of the more advanced features of Algol68, he thought it too complex, when he designed Pascal
Implementing correctly disjoint unions, i.e. allowing them to be used only exactly like an enumeration in the variable tested by a select/case/switch statement, and then only as the correct type in each of the alternatives, introduces a little overhead in a compiler, but it is certainly not a serious source of bloat in comparison with most other things that must be done by a compiler.
If the programming language designer had clearly in mind how disjoint union types must work, they would have been easy to implement even for the minicomputers and microcomputers of the seventies.
Even Go v1.0 type system is more advanced than Oberon-07 final form.
In my opinion, the most important contributions of Niklaus Wirth to programming languages have been in his early work on Euler, PL360 and ALGOL W, which introduced various features that were novel at that time.
Starting with Pascal in 1970, his programming languages remained reasonably good for teaching programming and how to implement a compiler, due to their simplicity, but all of them were seriously behind contemporaneous languages.
While Mesa of Xerox was a nice and innovative language, that cannot be said about Modula, Wirth's attempt to reimplement similar features after his sabbatical at Xerox, which was only mediocre.
On the other hand the early languages of Wirth were very innovative, e.g. Euler was one of the first 2 languages with pointers, the other being CPL. In contrast with CPL which had implicit pointer dereferencing, Euler had explicit address-of and indirection operators, but it got their syntax right, not like in C, where the indirection operator has been mistakenly defined as prefix, instead of postfix.
Zig is basically Modula-2 type system in C clothing, plus comptime, if only people were equally as hyped back in the 1980's.
The usual "Why Pascal..." fails flat in the presence of Modula-2, which was actually designed as systems language, and not as a language to learn about programming.
Oberon (the 1992 original), for its simplicity, introduced a wider audience to the concept that system programming with automatic resource management isn't something out of this world, even though Cedar is more interesting in features.
I was more interested into Component Pascal and Active Oberon, even though there were the work of other researchers at ETHZ.
Nonetheless it was his work, that inspired me to dig into everything Xerox PARC was doing, and discovering there was more happening there than only Smalltalk.
I became amazed at the work done across Interlisp-D, Mesa, Cedar, how advanced their ideas for what an IDE is supposed to look like, that many mainstream languages still can't offer.
So in a sense that was also a contribution from Niklaus Wirth to everyone that got interested into his work, and decided to go down the rabbit hole.