i would say when you use a.b!c.d couple of times in your code but dont feel like making a macro or setting it to a variable. Naming things is one of the hardest problems in CS so name like a.b!c.d suits that expression well as it turns out to be that expression.
that is not a bad idea, I was against adding infix stuff to lisp till i saw this this makes 1 still act like lisp and 2 look nice without any sort of additional syntax, good job eds.
Which 1 and 2 are you referring to exactly? The only points I see right now are those in the previous discussion but those don't entirely make sense in view of your comments.
I was really glad when cadaver suggested this solution to infix syntax. I just don't like the look of expressions like
+[*[2; f[x/3]]; y]
even if your syntax allows you to shorten math expressions to (for example)
2*f[x/3]+y
because in my opinion you lose on everything else. But now I'm just rambling....
In this particular case libraries don't matter much as the tasks in benchmarks are mostly algorithmical, e.g. like finding spectral norm of matrix, summing numbers or solving n-body problem. Maybe for development of language it would be great if somebody could write some benchmarks to test arc brevity.
