http://qs321.pair.com?node_id=791932


in reply to Re^5: Unparseability is A Good Thing
in thread Unparseability is A Good Thing

I'll have to get back to you on the first paragraph. The 5 minutes I have aren't enough

Please keep reading about non-determinism.

Any suggested reference?

I thought you were saying perl was not parsable.

I never said that. I know that Perl cannot be parsed without executing arbitrary Perl code, but I never even said that I believed that. I just answered your questions.

Replies are listed 'Best First'.
Re^7: Unparseability is A Good Thing
by Zen (Deacon) on Aug 28, 2009 at 15:46 UTC
    This is true of any language, though. Consider a function pointer in C. You can prototype all day long, but what if your prototype leads to a function pointer that gets borked partway through and you get a dump? What about NullPointerException in Java? This is all halting problem material, too. This isn't a perl issue. Part of the turing machine in models of computation is the tape and the read head; nowhere does it say that the Turing machine is omniscient of the entire input stream.

    Am I the only one unimpressed?

    As for reference material, I often wrongly assume everyone went through comp sci. I did undergrad and grad modcomp from two different- but now out of print- textbooks. There are different books now, such as the one by Michael Sipser (Introduction to the Theory of Computation) that should cover computability theory, turing machines, and automata.

    If it matters any, I remember /facepalm'ing many times through both of those courses at how obvious and unimpressive it was, just formalizations and proofs of what you already (probably) know.
[reply]

      but what if your prototype leads to a function pointer that gets borked partway through

      Sounds like you're talking about bugs, resource problems and other real-world considerations. Those are implementation issues, not algorithmic issues, and determinism is a property of algorithms.

      So yes, those are irrelevant. I never said they were.

      As for reference material, I often wrongly assume everyone went through comp sci.

      I have a B.CS. The course in which this was taught had a very boring teacher. I needed a booster.

[reply]
        Bugs, yes, but also parseability problems by your definition. If the parser cannot be expected to compute bad programming in the C example due to execution details, how can it be expected to in perl or any language?

        I am convinced all mod comp courses are snoozefests. It's like being taught the identity property in middle school and gritting your teeth at how obvious it all is. Yet the textbook costs over $100- people get paid for this!
[reply]

In Section Meditations