Complexity nomenclature

Andrei Alexandrescu via Digitalmars-d digitalmars-d at puremagic.com
Mon Dec 7 05:46:12 PST 2015


On 12/7/15 5:14 AM, Timon Gehr wrote:
> D does not allow overloading of syntax to the extent necessary to make
> similar things really pleasant in the long run, and it has been
> repeatedly argued that this is a good thing; that custom parsing should
> be used instead. It is easy to align the parser with D syntax. Anyway,
> syntax is not the problem here, and the implementation can be downgraded
> to not support parsing and/or proper names at any point.

I fail to see how no parens after log or "^" in lieu "^^" would make a 
positive difference. What would be a few examples of things that won't 
work pleasantly enough?

I'm not sure whether the DSL argument is well applied here. Clearly 
using D expressions for e.g. regex or SQL syntax would be at best 
avoided in favor of a DSL. In this case we're defining an algebra over 
restricted expressions, which are a subset of the usual mathematical 
expressions that D's syntax already emulates.

Looks like a debate on whether to choose one standard language vs. an 
obscure one (in this case invented ad-hoc) is starting. This is deja vu 
all over again :o).

I hope you won't mind if I give your idea a slightly different angle.


Andrei



More information about the Digitalmars-d mailing list