Thanks, Tim. The reason I can't run specific lines from the test file is
that the test needs to call module B first to get something which will be
passed as argument to a function in module A and it is this function that
I'm testing. Here is what it looks like:
Module A:
module A
import B
function f(p)
...
B.g()
...
end
end
Module B:
module B
function g()
...
return p
end
end
test file:
import A, B
p = B.g()
A.f(p)
I want to see how many times B.g() is called by A only or how much the
computation of B.g() is triggered by module A/A.f() only, instead of by A
and the test. Hope that I have made my point clear.
On Tuesday, July 5, 2016 at 3:43:16 PM UTC+1, Tim Holy wrote:
>
> Ah. Can't you just run specific lines from that test file? You could even
> copy
> it and then delete the irrelevant lines using an editor, if you need to
> run
> many tests.
>
> --Tim
>
> On Tuesday, July 5, 2016 7:03:07 AM CDT [email protected] <javascript:>
> wrote:
> > Tim, Thanks.
> >
> > The problem is this: I wrote a module A and made a test to test module
> A.
> > But both of module A and the test itself need to call another module B
> > frequently. I want to see how A is calling B and don't care how the test
> is
> > calling B. That's why I want to separate the profiling of module A from
> > that of both. This way I can easily the profiling via
> >
> > julia> Profile.print(format=:flat, sortedby=:count)
> >
> > Sorry for my dumb question.
> >
> > On Tuesday, July 5, 2016 at 12:49:37 PM UTC+1, Tim Holy wrote:
> > > I don't know what "profiling a module" means. You profile running
> code,
> > > wherever
> > > that code happens to live---and that's all there is to say. To profile
> the
> > > code
> > > in a module, you just have to write code that exercises the functions
> in
> > > the
> > > module.
> > >
> > > The meaning of the numbers is described here:
> > > http://docs.julialang.org/en/stable/manual/profile/#basic-usage
> > > The key words are "sampling profiler," the meaning of which is
> described
> > > at the
> > > top of that page (and see the wikipedia link). The number of samples
> is
> > > approximately proportional to the cost of the line (or its
> descendents).
> > >
> > > Best,
> > > --Tim
> > >
> > > On Tuesday, July 5, 2016 4:40:32 AM CDT [email protected]
> <javascript:>
> > >
> > > wrote:
> > > > Bump up.
> > > >
> > > > On Monday, July 4, 2016 at 4:33:53 PM UTC+1, [email protected]
> wrote:
> > > > > I want to profile a module which is tested by a test. Following
> the
> > > > > documentation (
> > >
> > >
> http://docs.julialang.org/en/release-0.4/manual/profile/#options-for-contr
> > >
> > > > > olling-the-display-of-profile-results), I know how to profile them
> > >
> > > (module
> > >
> > > > > + test) together:
> > > > >
> > > > > @profile include("test.jl")
> > > > >
> > > > >
> > > > > But I have no idea for how to do the profiling only for the
> module.
> > > > >
> > > > > A second question is about the first number of each line in
> profiler's
> > > > > output. For example, the output from Julia Documentation:
> > > > >
> > > > > julia> Profile.print()
> > > > >
> > > > > 23 client.jl; _start; line: 373
> > > > >
> > > > > 23 client.jl; run_repl; line: 166
> > > > >
> > > > > 23 client.jl; eval_user_input; line: 91
> > > > >
> > > > > 23 profile.jl; anonymous; line: 14
> > > > >
> > > > > 8 none; myfunc; line: 2
> > > > >
> > > > > 8 dSFMT.jl; dsfmt_gv_fill_array_close_open!;
> line:
> > > 128
> > >
> > > > > 15 none; myfunc; line: 3
> > > > >
> > > > > 2 reduce.jl; max; line: 35
> > > > > 2 reduce.jl; max; line: 36
> > > > > 11 reduce.jl; max; line: 37
> > > > >
> > > > > Is it appropriate to interpret the numbers 23, 8, 15, etc as the
> > >
> > > number of
> > >
> > > > > times the line is run or the time has been spent (relatively) on
> that
> > > > > line?
> > > > >
> > > > >
> > > > > I searched the group and there is no threads with a similar topic.
> Any
> > > > > comments? Thanks!
>
>
>
