On 13.01.2013, at 10:59, Ross Bencina <rossb-lists_at_audiomulch.com> wrote:
>
>
> On 13/01/2013 8:17 AM, David Barbour wrote:
>>
>> In general no. Most user interfaces are not, in any formal sense,
>> programming languages.
>> To be a programming language you need Turing completeness.
>>
>> Oh? Eight years studying PL and that's the first I've heard of this
>> significant restriction. I better go tell the Agda, Idris, Coq, Funloft,
>> Synchrone, and Charity people that they've got it all wrong.
>>
>> But them SQL guys got it right. SQL's been Turing complete since 1998.
>> Oh, and it seems CSS+HTML is also Turing complete (even without JavaScript).
>
> I don't really care whether researchers want to call sub-turing-complete functions "programming languages." If you don't have universal computation then it's not a programming language in my book. This is pretty straight forward and clear cut.
>
> Further, if your program/function is decidable it's not very interesting from a performance standpoint.
>
> Clearly I'm in the minority, but the line has to be drawn somewhere and I can't think of a better place to draw it -- unless you want some watered down definition of "live coding" that pretty much admits every procedural activity under the sun from deterministic dataflow patching to musical dice games.
Ross, I think the issue is the difference between absolute universality (Turing Completeness) and a type of relative universality or intractability, which makes also those systems interesting which are simpler (e.g. finite state machines). There is a lot of structure in the space of unreachable but decidable problems. I share your intuition, however, that there is a certain quality of formality, a quality that makes programming transcend a series of specifications. But the issue is a little more intricate and has more to do with how delegation of thought really works. I find it important that the activities of programming can be delegated to the program, as well as that activities of the program can be delegated back to the programmer. But this is not a criterion. The totality of the live coding situation is dependent on a lot of factors which are not in the system structure alone. Arguably, this is the same for all language situations, including those involving computer languages. For instance, in analogy you may ask "what is necessary for a conversation proper?" or "What is a thought?".
Julian
Received on Sun Jan 13 2013 - 15:54:33 GMT