Add explanation about top-level compilation to timing macros

[IanButterworth]

Following discussions in #39760

Text based on @JeffBezanson's edits in JuliaLang/www.julialang.org@ef6846d

[IanButterworth]

To illustrate the need for this note

julia> @eval function f(a)  # Expensive function to compile. (Thanks @yuyichao!)
                         $(quote $([:(a = a + a * 10 + (a > 0 ? $i : 0.2)) for i in 1:401]...) end)
                         $(quote $([:(a = a + a * 10 + (a > 0 ? $i : 0.2)) for i in 1:401]...) end)
                         return a
                     end
f (generic function with 1 method)

julia> @time f(2)
  0.000000 seconds

julia> @eval function f(a)  # Expensive function to compile. (Thanks @yuyichao!)
                         $(quote $([:(a = a + a * 10 + (a > 0 ? $i : 0.2)) for i in 1:401]...) end)
                         $(quote $([:(a = a + a * 10 + (a > 0 ? $i : 0.2)) for i in 1:401]...) end)
                         return a
                     end
f (generic function with 1 method)

julia> @time @eval f(2)
 13.801006 seconds (6.26 M allocations: 317.365 MiB, 3.65% gc time, 100.02% compilation time)

[Keno]

I wonder if there is some way that we can have the former expression hint that compile time was excluded. I guess we could have the REPL check if what you're evaling is a literal @time ... and give you a hint, but that might be a little out there. I'm just a big fan of in-context discovery, since I basically never read the docs ;)

[ufechner7]

Looks good to me. :)

[ufechner7]

What is the reason for the failure on linux32? Is this ready to be merged?

[ufechner7]

Bump.

[DilumAluthge]

@JeffBezanson @Keno What are your thoughts on this PR?

[ufechner7]

Can this PR get a backport label? It would be nice to have it in 1.6 .