Talk:Main Page

From Successor ML

Providing encouragement for

  'functional record update'
  'punning'

not that I am discouraging any of the ideas so far. However, with regard to introducing new keywords, please keep Ockham's razor in mind. PL/I did not, with 255 non-reserved keywords. While I suppose some still use it, it is 'dead' in Bob Harper's sense. Even c++ has a disturbingly large number of keywords, a reflection on the heavy constraints that went into that language's design.

Similarly with regard to new syntax. Here I am hoping for something better with regard to functional record update, not wanting to have to try to explain the role of { and } in this setting. For beginners, at least, using ordinary parentheses for grouping will be clearer.

Dbenson@eecs.wsu.edu 17:07, 29 October 2005 (EDT)

Deadness

Standard ML, being incapable of evolution, is, therefore, dead. The purpose of successor ML, or sML for short, is to provide a vehicle for the continued evolution of ML, using Standard ML as a starting point. The intention is for successor ML to be a living, evolving dialect of ML that is responsive to community needs and advances in language design, implementation, and semantics.

This is an interesting statement. Isn't there already a "living, evolving dialect of ML that is responsive to community needs" called Ocaml? For some of us, the attraction of SML has been precisely its stability coupled with the formal definition. Otherwise we would've used Ocaml which has some definite practical advantages. Can a "living, evolving dialect of ML that is responsive to community needs"

1/ still be formally defined in its entirety, with the definition kept in 1 document (i.e. not like Haskell which has formal definition for most - or all - of the new bits, but it is scattered in dozens of research papers)

2/ have multiple implementations differing in tools, optimization emphasis etc. but implementing precisely the same language?

I doubt it. See Haskell.

[please sign your comments.]

On Doubt

I think that the right answer is to try and strike a balance between stability and (slow) improvements. Also, with respect to point 1 above we need to make sure that whatever change is being made is made by way of providing the appropriate formal definition within the one document defining the language in its entirety. This would also facilitate point 2 (although there will be times when one or the other implementation lags behind in implementing the full feature set).

Of course, development having been stagnant for such a long time, Successor ML is due for some relatively rapid changes in the beginning. We should start with fixing obvious problems and getting the low-hanging fruit.

Blume 14:22, 15 December 2005 (EST)

Permissions?

Does anyone know what kind of permissions Milner and Tofte etc might give for the Definition of Standard ML? I am guessing that the rules themselves could probably just be used since they are basically mathematical equations, but copying the full text would require permission. This kind of permission would make this project much easier. Jrincayc 03:35, 10 February 2010 (UTC)