Re: [Caml-list] Hitchhiker's Guide to Typing

[Brian Rogoff <bpr@best.com>]

On Fri, 8 Jun 2001 leary@nwlink.com wrote:
> On Fri, Jun 08, 2001 at 01:22:05PM -0700, Hao-yang Wang wrote:
> > I think by "type theory" you mean the type systems used in modern 
> > programming languages. Luca Cardelli has written some nice 
> > tutorial/survey on this topic.
> > 
> > See <http://www.luca.demon.co.uk/Bibliography.html>. Look for the 
> > articles "Typeful programming" and "Type systems".

If, like me, you like to see a working implementation, then download a 
copy of "Basic Polymorphic Typechecking" from this page. Also, Michael 
Schwartzbach wrote a nice little tutorial called "Polymorphic Type
Inference" which you can get here 

http://www.brics.aau.dk/BRICS/LS/95/3/BRICS-LS-95-3/BRICS-LS-95-3.html

There are little parts of this that I think could be better, but overall 
I like it best. If you work through it you'll learn a lot. 

> Seems to beg the questions tho': Is it not possible to learn and use OCaml
> without wading through 60 page docs on typing? 

Sure. But you'll understand these things a lot better if you apply
yourself and read some of these papers. I agree that many of the papers 
go too deep too fast. 

> Will I have to read more of these papers to learn the object
> system?  Modules?

No, but once again you'll learn a lot from the papers. For objects and
modules 

http://caml.inria.fr/~remy/cours/appsem/
http://caml.inria.fr/~xleroy/publi/modular-modules-jfp.ps.gz

> I'm willing to read up on typing, since it seems pretty important to
> getting a good handle on FPLs/OCaml, and that's definitely what I'm after,
> but I wonder (in advance) if 60 pages can't be turned into far fewer.  I'd
> like some feedback on the notion that "if you can't explain it to a
> five-year-old, you don't really understand it"

That notion is nonsense. I understand differential calculus quite well,
and I doubt I could explain it to a five year old. In fact, as a TA for  
an introductory numerical analysis class at Snodfart (name changed to
protect the guilty :) I gave a simple explanation of Taylor's Theorem to a 
masters student in CS who just couldn't get it. Not a very good student,
no doubt, but probably better equipped than a five year old. 

> (or, rather, haven't given the simplest, most concise, most practically
> useful explanation).  By way of a couple examples, the HHGTTG entry for
> Earth is "mostly harmless". Investing can be distilled down to "buy
> low, sell high".  Not crashing a motorcycle is a matter of "look where
> you want to go, don't look at the ground".

None of these are useful. My explanation for everything can be distilled
to "Shit happens", but that's also not very useful.

> Of course, a working knowledge of typing will never be so simple, but
> contrast, for example, the "Type Systems" doc above; Table 34 is a good
> place to look.

I've got a copy of Stroustrup's "The C++ Programming Language" here. That's 
not simple. Nor is Wall's Perl book, nor the Common Lisp Hyperspec, nor
the Ada 95 LRM. 

If you want a working knowledge of OCaml, write lots of code. Use the
papers to flesh out that working knowledge to a deeper understanding. 

-- Brian


-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr