On Thu, 4 May 2023 at 00:30, Gaius Mulley wrote:
ah yes I recall the Milton Keynes meeting and the desire to see Modula-2
ported to some ancient hardware with a restricted ASCII set. I vaguely
recall the consternation in the room over BCD and type COMPLEX -
although this might be an over active imagination on my part :-).
Your recollection is correct.
The consternation over BCD and COMPLEX was the fault of myself and Albert Wiedemann as we had tabled a proposal to add a native BCD type to Modula-2 so that infix notation could be used in arithmetic operations on type BCD. But this was seen as a pandora's box and we were told that if we were to get our BCD type, then Rick Sutcliffe would get his COMPLEX type and sooner or later somebody else wanted yet another and another.
To the applause of the room, but to Rick's disappointment, Albert and I then decided to withdraw our proposal, Many years later, when most participants had lost interest, Rick made another approach and got his COMPLEX type then. And Albert implemented a native BCD type as an extension in his p1 compiler.
Yes indeed all languages use operator overload to an extent - I can see
why it moved into user space
There is a middle ground though: Permit libraries to define functions to 'overload' *existing* syntax/operators, but not to create new syntax/operators. That's the route we've taken in R10. We don't call it overloading though, we call it syntax binding. BTW, Wirth did this in PIM with NEW and DISPOSE.
regards
benjamin