I remember, back in 1985, trying to write a parser in Prolog on a VAX 11/750. I didn't know what I was doing and my code fell into the exponential case trap. Even minimalist examples appeared to crash as they took unreasonably long, so I wasn't getting clues to the problem, and just gave up.
Consequently I want to rephrase the first sentence of your third paragraph. The problem with DCGs is performance: they're exponential in the naive case. That might not sound too bad today, but back in 1985 the hype was that Prolog let you program declaratively. Just declare what a parse looks like. Reasonable performance in the naive case was the popular selling point for Prolog.
Comments
I remember, back in 1985, trying to write a parser in Prolog on a VAX 11/750. I didn't know what I was doing and my code fell into the exponential case trap. Even minimalist examples appeared to crash as they took unreasonably long, so I wasn't getting clues to the problem, and just gave up.
Consequently I want to rephrase the first sentence of your third paragraph. The problem with DCGs is performance: they're exponential in the naive case. That might not sound too bad today, but back in 1985 the hype was that Prolog let you program declaratively. Just declare what a parse looks like. Reasonable performance in the naive case was the popular selling point for Prolog.