> this is awesome
>
> thanks!
A small word of caution: macros can make it harder for others (or your
future self) to read the code, as essentially each macro introduces a
new keyword. Just something to consider when introducing a macro;
awesome as they are.
> On Wednesday, February 24, 2016 at 3:32:48 PM UTC+1, Tom Breloff wrote:
>>
>> Enjoy:
>>
>>
>> julia> macro n(expr::Expr)
>> @assert expr.head == :function
>> s = expr.args[1].args[2]
>> if isa(s,Expr) && s.head == :(::)
>> s = s.args[1]
>> end
>> unshift!(expr.args[2].args, :(n = length($s)))
>> esc(expr)
>> end
>>
>> julia> macroexpand(:( @n function f(x,y)
>> println("length of x is $n")
>> end))
>> :(function f(x,y)
>> n = length(x) # none, line 2:
>> println("length of x is $(n)")
>> end)
>>
>> julia> @n function f(x,y)
>> println("length of x is $n")
>> end
>> f (generic function with 1 method
>>
>> julia> @n function f(x::Vector{Int},y)
>> println("INTVEC: length of x is $n")
>> end
>> f (generic function with 2 methods)
>>
>> julia> f(rand(4),rand(5))
>> length of x is 4
>>
>> julia> f(rand(Int,4),rand(5))
>> INTVEC: length of x is 4
>>
>>
>> On Wed, Feb 24, 2016 at 9:03 AM, Alex <[email protected] <javascript:>>
>> wrote:
>>
>>> Hi everybody,
>>>
>>> I have several functions operating on one dimensional vectors.
>>>
>>> Within each function the first thing I do is to define variable n holding
>>> the length of input vector x:
>>>
>>> n = length(x)
>>>
>>> Is that possible to factor out this pattern somehow from all functions
>>> definitions?
>>>
>>> (...like 'traits' in Scala)
>>>
>>> I mean, to make variable n automatically defined as a length of the first
>>> argument (which is always a one dimensional vector) of each function?
>>>
>>>
>>>
>>
