On Wednesday, August 5, 2015 at 10:59:18 PM UTC-4, Tim Holy wrote:
>
> You may want to check the IProfile.jl package, which does a lot of this
> kind of
> stuff (including inserting timing code around each line of a function).
>
That was an interesting read, thank you. As a common lisper, all those
gensyms were oddly comforting.
Cédric
>
> --Tim
>
> On Wednesday, August 05, 2015 06:47:31 PM Cedric St-Jean wrote:
> > Yichao Yu: How would you retrieve the counter's value? How would you
> > distinguish between the counter in function foo vs. the one in function
> bar?
> > On Wednesday, August 5, 2015 at 3:56:12 PM UTC-4, Yichao Yu wrote:
> > > On Wed, Aug 5, 2015 at 2:42 PM, Cedric St-Jean <cedric...@gmail.com
> > >
> > > <javascript:>> wrote:
> > > > In my case, I wanted to count how many times some functions were
> called.
> > > > Kinda like a profiler, but for a different purpose. If the function
> name
> > >
> > > was
> > >
> > > > available, I could move the macro "call" inside the function body,
> which
> > > > would make more sense.
> > >
> > > Than what about sth like
> > >
> > > ```julia
> > > macro count_execute()
> > >
> > > @gensym counter
> > > quote
> > >
> > > global $counter
> > > if isdefined(:($counter))
> > >
> > > $counter = $counter::Int + 1
> > >
> > > else
> > >
> > > $counter = 0
> > >
> > > end
> > >
> > > end
> > >
> > > end
> > > ```
> > >
> > > > On Wednesday, August 5, 2015 at 2:25:42 PM UTC-4, Isaiah wrote:
> > > >> I'm still curious what problem / design this is trying to solve. Be
> > >
> > > aware
> > >
> > > >> that doing something like this for custom runtime dispatching is
> likely
> > >
> > > to
> > >
> > > >> be slow (except in the macro case which does not really count).
> > > >>
> > > >> Are you both coming from some other programming language that
> exposes
> > >
> > > this
> > >
> > > >> in a first-class way? Both PHP and JavaScript seem to make this
> > >
> > > information
> > >
> > > >> available, but R does not (except for a callstack hack).
> > > >>
> > > >> On Wed, Aug 5, 2015 at 2:10 PM, Cedric St-Jean <cedric...@gmail.com>
>
> > > >>
> > > >> wrote:
> > > >>> I also wish that the name was visible... I wrote the macro a few
> days
> > > >>> ago:
> > > >>>
> > > >>> macro expose_name(fdef)
> > > >>>
> > > >>> @capture(fdef, begin function fname_ (args__) body__ end end)
> > > >>> esc(:(function $(fname) ($(args...))
> > > >>>
> > > >>> let function_name = $(Expr(:quote, fname))
> > > >>>
> > > >>> $(body...)
> > > >>>
> > > >>> end))
> > > >>>
> > > >>> end
> > > >>>
> > > >>> @expose_name function foo () @show function_name end
> > > >>>
> > > >>> Cédric
> > > >>>
> > > >>> On Wednesday, August 5, 2015 at 12:54:38 PM UTC-4, Tim Holy wrote:
> > > >>>> Alternatively,
> > > >>>>
> > > >>>> julia> for (offset, f) in enumerate((:foo1, :foo2, :foo3))
> > > >>>>
> > > >>>> @eval begin
> > > >>>>
> > > >>>> function $f(x)
> > > >>>>
> > > >>>> println("This is ", $f)
> > > >>>> x + $offset
> > > >>>>
> > > >>>> end
> > > >>>>
> > > >>>> end
> > > >>>>
> > > >>>> end
> > > >>>>
> > > >>>> julia> foo1(0)
> > > >>>> This is foo1
> > > >>>> 1
> > > >>>>
> > > >>>> julia> foo2(0)
> > > >>>> This is foo2
> > > >>>> 2
> > > >>>>
> > > >>>> julia> foo3(0)
> > > >>>> This is foo3
> > > >>>> 3
> > > >>>>
> > > >>>> --Tim
> > > >>>>
> > > >>>> On Wednesday, August 05, 2015 12:48:29 PM Isaiah Norton wrote:
> > > >>>> > Can I ask what is your use-case, and whether you've done this
> in
> > >
> > > some
> > >
> > > >>>> > other
> > > >>>> > language? From a quick google there are appear to be some
> (possibly
> > > >>>> > non-standard) C and C++ compiler defines that provide the
> current
> > > >>>> > function
> > > >>>> > name, but other than that:
> > > >>>> >
> > > >>>> > - Python rejected a PEP for this feature (
> > > >>>> > https://www.python.org/dev/peps/pep-3130/)
> > > >>>> > - Java does not support it in a first-class way ("could use
> > > >>>> > stacktrace, but
> > > >>>> > potentially unreliable")
> > > >>>> > - some lisps appear to have interactive mode support for this
> kind
> > >
> > > of
> > >
> > > >>>> > introspection, but nothing general.
> > > >>>> >
> > > >>>> > The recommended way to solve problems that require this level
> of
> > > >>>> > introspection is to do code generation via metaprogramming.
> > > >>>> >
> > > >>>> >
> > > >>>> > On Wed, Aug 5, 2015 at 12:36 PM, Isaiah Norton
> > > >>>> > <isaiah...@gmail.com>
> > > >>>> >
> > > >>>> > wrote:
> > > >>>> > > No.
> > > >>>> > >
> > > >>>> > > Ok, technically, you could do this (inspired by
> > >
> > > https://github.com/JuliaLang/julia/issues/8066#issuecomment-61136584):
>
> > > >>>> > > julia> function foo()
> > > >>>> > >
> > > >>>> > > bt = backtrace()
> > > >>>> > > lookup = ccall(:jl_lookup_code_address, Any,
> (Ptr{Void},
> > > >>>> > >
> > > >>>> > > Int32),
> > > >>>> > >
> > > >>>> > > bt[2], 0)
> > > >>>> > >
> > > >>>> > > name = lookup[1]
> > > >>>> > > end
> > > >>>> > >
> > > >>>> > > foo (generic function with 1 method)
> > > >>>> > >
> > > >>>> > > julia> foo()
> > > >>>> > >
> > > >>>> > > :foo
> > > >>>> > >
> > > >>>> > > But that is a really, really bad idea. Please don't do that.
> > > >>>> > >
> > > >>>> > > On Wed, Aug 5, 2015 at 12:04 PM, Dominique Orban <
> > > >>>> > >
> > > >>>> > > dominiq...@gmail.com> wrote:
> > > >>>> > >> Sorry if this has been asked before. Is it possible to
> determine
> > > >>>> > >> the name
> > > >>>> > >> of a function inside that function? For example,
> > > >>>> > >>
> > > >>>> > >> function blah(x)
> > > >>>> > >>
> > > >>>> > >> my_name = ... # should evaluate to "blah" or :blah
> > > >>>> > >>
> > > >>>> > >> end
> > > >>>> > >>
> > > >>>> > >> I didn't see that in the introspection section of the
> > > >>>> > >> documentation.
> > > >>>> > >>
> > > >>>> > >> Thanks!
>
>
