Newbyish question about Array slice method

I bought Crockford's "javascript: The Good Parts" yesterday to help
build my JavaScript foo.

On page 44, he gives an implementation of the curry function:

Function.method('curry', function() {
var slice = Array.prototype.slice,
args = slice.apply(arguments),
that = this;

return function() {
return that.apply(null, args.concat(slice.apply(arguments)));
};
});

I'm trying to reconcile the two invocations of slice.apply(arguments)
with the notion that the slice method in Array's prototype has two
arguments, the second of which is optional.

I thought that this would mean that those invocations would need to be
slice.apply(arguments,0) in order to supply the start argument, and in
"leap before you look" mode I submitted an erratum to O'Reilly.

Now however, I've actually tried the code in both Firefox and Safari,
and it seems to work as it's given in the book.

So is the first argument to slice really optional despite the various
doc's I've checked (including and besides the book in question), or
are Firefox and Safari both working outside of the JS spec here, or is
there something else I'm missing?

On Jun 1, 9:52 pm, RubyRedRick <rick.denat... (AT) gmail (DOT) com> wrote:

So is the first argument to slice really optional despite the various
doc's I've checked (including and besides the book in question), or
are Firefox and Safari both working outside of the JS spec here, or is
there something else I'm missing?

The docs are wrong: at least for an engine strictly implementing the
relevant ECMA-262 internal algorithms.

Array slice method production:

15.4.4.10
....
4. Call ToInteger(start).
5. If Result(4) is negative, use max((Result(3)+Result(4)),0); else
use min(Result(4),Result(3)).

Internal ToInteger production:

9.4 ToInteger
1. Call ToNumber on the input argument.
2. If Result(1) is NaN, return +0.
....

This way slice() without arguments still will equal to slice(0) so
still be working. This is what I don't like rather often in
Crockford's approaches: they are bearing to much of mannerism for my
blood, to much of coding just for coding. Respectively sometimes
_unnecessary_ dependence on strict engine implementations. Does it
kill anyone to provide the argument? Why to transform a business
solution into some kind of engine's acid test? It is strictly my
opinion and the book contains an ocean of interesting approaches. Just
be warned that it is not a book I would recomment to anyone to learn
javascript: it is for advanced programmers willing to improve their
skills. Again strictly IMHO.

RubyRedRick wrote:

I bought Crockford's "javascript: The Good Parts" yesterday
to help build my JavaScript foo.

On page 44, he gives an implementation of the curry function:

Function.method('curry', function() {
var slice = Array.prototype.slice,
args = slice.apply(arguments),
that = this;

return function() {
return that.apply(null, args.concat(slice.apply(arguments)));
};
});

I'm trying to reconcile the two invocations of
slice.apply(arguments) with the notion that the slice method in
Array's prototype has two arguments, the second of which is
optional.

I thought that this would mean that those invocations would need
to be slice.apply(arguments,0) in order to supply the start
argument, and in "leap before you look" mode I submitted an
erratum to O'Reilly.

Now however, I've actually tried the code in both Firefox and
Safari, and it seems to work as it's given in the book.

So is the first argument to slice really optional despite the
various doc's I've checked (including and besides the book in
question), or are Firefox and Safari both working outside of
the JS spec here, or is there something else I'm missing?

Perhaps you are falling to look at the JS Spec itself (ECMA 262, 3rd
Ed.)? the forth step in the algorithm for - Array.prototype.slice - is
"Call ToInteger(start)", and the internal ToInteger function will return
numeric zero when its input is the undefined value (which is what its
input is going to be when the argument is not passed in).

A much more interesting question is why use - apply - when you could
use - call -? Not that the former is a mistake (the handling of the
undefined second argument is fully specified) but the - apply - method
has a length of 2 and the call method has a length of 1, which suggests
it should be the one to use when there is only going to be one argument.

Richard.

On Jun 1, 2:37*pm, "Richard Cornford" <Rich... (AT) litotes (DOT) demon.co.uk>
wrote:

RubyRedRick wrote:
I bought Crockford's "javascript: The Good Parts" yesterday
to help build my JavaScript foo.

On page 44, he gives an implementation of the curry function:

Function.method('curry', function() {
var slice = Array.prototype.slice,
args = slice.apply(arguments),
that = this;

*return function() {
* *return that.apply(null, args.concat(slice.apply(arguments)));
};
});


A much more interesting question is why use - apply - when you could
use - call -? Not that the former is a mistake (the handling of the
undefined second argument is fully specified) but the - apply - method
has a length of 2 and the call method has a length of 1, which suggests
it should be the one to use when there is only going to be one argument.

I don't see how call could be used here. The method is being used as
a trick to obtain a real array object from arguments which is a pseudo-
array and lacks methods.

RubyRedRick wrote:

"Richard Cornford" wrote:
RubyRedRick wrote:
I bought Crockford's "javascript: The Good Parts" yesterday
to help build my JavaScript foo.
On page 44, he gives an implementation of the curry function:
Function.method('curry', function() {
var slice = Array.prototype.slice,
args = slice.apply(arguments),
[1]--------------^^^^^

that = this;
return function() {
return that.apply(null, args.concat(slice.apply(arguments)));
[2]----------------------------------------------^^^^^

};
});

A much more interesting question is why use - apply - when you could
use - call -? Not that the former is a mistake (the handling of the
undefined second argument is fully specified) but the - apply - method
has a length of 2 and the call method has a length of 1, which suggests
it should be the one to use when there is only going to be one argument.

I don't see how call could be used here. The method is being used as
a trick to obtain a real array object from arguments which is a pseudo-
array and lacks methods.

You can use Function.prototype.call() instead of Function.prototype.apply()
at [^1] and [^2] because there is no second argument there.


PointedEars