True arity

Is there a way to get the maximum number of arguments that a method
can receive? Method#arity only gives the lower bound…

Daniel

On 5/29/07, Daniel DeLorme [email protected] wrote:

Is there a way to get the maximum number of arguments that a method
can receive? Method#arity only gives the lower bound…

I am afraid I fail to understand, what would you define as the maximum?

def a(a)… I would say the maximum is 1 which is the arity
def b(*b) arity is -1 but what would the maximum of parameters be?
def c(c,*d) arity = -2 but some question as above.

The question is, what would you like to achieve?

Cheers
Robert

Robert D. wrote:

The question is, what would you like to achieve?
In the last 2 cases you give the maximum would be Infinity. But in a
case like this:
def foo(a, b=nil, c=nil)
the maximum would be 3 but arity only gives me -2 which means “this
method has 1 required argument”. If I invoke the method like:
foo(*args)
I want to know for what size of args is this valid? It would be nice if
arity returned a Range object (1…3)

Daniel

arity is not able to show you all the information you need, default
parameter are a curse in this regard (or the format arity uses - a
simple integer is not enough to show all cases… even a range
wouldn’t, since
(1…(1.0/0.0))

1…Infinity

won’t tell you that much (i think)
I don’t know of a really simple programmatic way to determine the
arity, ruby2ruby and manual parsing might help you there if you
really really need it. Best would be to just raise and tell the user
he used the wrong arity. Or you could do a rundown, passing as many
parameters as possible and going down until it doesn’t raise any more.

Please note that this post is full of bad practice and evil hacks :slight_smile:

^ manveru

On 5/29/07, Daniel DeLorme [email protected] wrote:

The question is, what would you like to achieve?

In the last 2 cases you give the maximum would be Infinity. But in a
case like this:
def foo(a, b=nil, c=nil)
I know I missed something :(, indeed a good question, thx for
explaining to a dummy ;).
the maximum would be 3 but arity only gives me -2 which means “this
method has 1 required argument”. If I invoke the method like:
foo(*args)
I want to know for what size of args is this valid? It would be nice if
arity returned a Range object (1…3)

Daniel

Cheers
Robert

Daniel DeLorme wrote:

In the last 2 cases you give the maximum would be Infinity. But in a
case like this:
def foo(a, b=nil, c=nil)
the maximum would be 3 but arity only gives me -2 which means “this
method has 1 required argument”. If I invoke the method like:
foo(*args)
I want to know for what size of args is this valid? It would be nice if
arity returned a Range object (1…3)

It could be done, but the current Ruby implementations don’t provide
this information. Of course, all of them could…we know how many
required arguments there are, we know how many optional arguments there
are, and we know if there’s a “rest” arg or not.

Maximum =

…base arity if no optional args or rest arg
…base arity plus optional arg count if no rest arg
…infinity if rest arg in any case

Minimum = …well you know it, because we have this today.

I guess the question would be whether this is useful enough information
to warrant a semantic change in “arity” representation.

  • Charlie

On 5/30/07, Daniel DeLorme [email protected] wrote:

IMHO, Range objects are underused. MatchData#offset would be so much
nicer if it returned (a…b) instead of [a,b]

just my 2¥
that is clever Daniel :wink:

but seriously I just wanted to backup your idea which seems pretty
good, what about an RCR?
And yes, by all means, use a range, look at this pretty idiom which
would result

raise ArgumentError unless method.arityrange === n

Cheers
Robert

Charles Oliver N. wrote:

I guess the question would be whether this is useful enough information
to warrant a semantic change in “arity” representation.

It can be useful in some cases and it is certainly cleaner than the
wondrous hack that is a negative arity, but I’m never very fond of
backwards compatibility breakage. I’d be more in favor of introducing a
new method (e.g. ‘arityrange’) that would return a Range object. (please
pretty please mr. core developpers)

IMHO, Range objects are underused. MatchData#offset would be so much
nicer if it returned (a…b) instead of [a,b]

just my 2¥

Daniel