deprecation warning is only printed once

[fredrikekre]

Ref #22034 (comment), on 0.6 and master I can only produce one deprecation warning per julia session.

master and 0.6-rc2:

julia> f2(x) = 2; f4(x) = 4;

julia> @deprecate f1(x) f2(x);

julia> @deprecate f3(x) f4(x);

julia> f1(1)
WARNING: f1(x) is deprecated, use f2(x) instead.
Stacktrace:
 [1] depwarn(::String, ::Symbol) at ./deprecated.jl:64
 [2] f1(::Int64) at ./deprecated.jl:51
 [3] eval(::Module, ::Any) at ./boot.jl:235
 [4] eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:66
 [5] macro expansion at ./REPL.jl:97 [inlined]
 [6] (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:73
while loading no file, in expression starting on line 0
2

julia> f3(1) # why does this not warn?
4

And on 0.5.2:

julia> f2(x) = 2; f4(x) = 4;

julia> @deprecate f1(x) f2(x);

julia> @deprecate f3(x) f4(x);

julia> f1(1)
WARNING: f1(x) is deprecated, use f2(x) instead.
 in depwarn(::String, ::Symbol) at ./deprecated.jl:64
 in f1(::Int64) at ./deprecated.jl:50
 in eval(::Module, ::Any) at ./boot.jl:234
 in eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:64
 in macro expansion at ./REPL.jl:95 [inlined]
 in (::Base.REPL.##3#4{Base.REPL.REPLBackend})() at ./event.jl:68
while loading no file, in expression starting on line 0
2

julia> f3(1)
WARNING: f3(x) is deprecated, use f4(x) instead.
 in depwarn(::String, ::Symbol) at ./deprecated.jl:64
 in f3(::Int64) at ./deprecated.jl:50
 in eval(::Module, ::Any) at ./boot.jl:234
 in eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:64
 in macro expansion at ./REPL.jl:95 [inlined]
 in (::Base.REPL.##3#4{Base.REPL.REPLBackend})() at ./event.jl:68
while loading no file, in expression starting on line 0
4

Seems like a bad bug in these days of new releases and fixing deprecation warnings.

[fredrikekre]

Bisected to c8026a9 and also ref #21367 (comment) which I also experienced writing tests for #22034

[fredrikekre]

Actually, its not just c8026a9, reverting that makes the warning print every time, so it has to be something else (perhaps a combination?). This is master with c8026a9 reverted:

julia> f2(x) = 2;

julia> @deprecate f1(x) f2(x);

julia> f1(1)
WARNING: f1(x) is deprecated, use f2(x) instead.
Stacktrace:
 [1] depwarn(::String, ::Symbol) at ./deprecated.jl:70
 [2] f1(::Int64) at ./deprecated.jl:57
 [3] eval(::Module, ::Any) at ./boot.jl:235
 [4] eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:66
 [5] macro expansion at ./REPL.jl:97 [inlined]
 [6] (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:73
while loading no file, in expression starting on line 0
2

julia> f1(1)
WARNING: f1(x) is deprecated, use f2(x) instead.
Stacktrace:
 [1] depwarn(::String, ::Symbol) at ./deprecated.jl:70
 [2] f1(::Int64) at ./deprecated.jl:57
 [3] eval(::Module, ::Any) at ./boot.jl:235
 [4] eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:66
 [5] macro expansion at ./REPL.jl:97 [inlined]
 [6] (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:73
while loading no file, in expression starting on line 0
2

[fredrikekre]

Perhaps this is only an issue in global scope. Example from top post, but calling f1 and f3 from a function, then it works as expected:

julia> f2(x) = 2; f4(x) = 4;

julia> @deprecate f1(x) f2(x);

julia> @deprecate f3(x) f4(x);

julia> function foo()
           f1(1)
           f3(1)
       end
foo (generic function with 1 method)

julia> foo()
WARNING: f1(x) is deprecated, use f2(x) instead.
Stacktrace:
 [1] depwarn(::String, ::Symbol) at ./deprecated.jl:70
 [2] f1(::Int64) at ./deprecated.jl:57
 [3] foo() at ./REPL[4]:2
 [4] eval(::Module, ::Any) at ./boot.jl:236
 [5] eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:66
 [6] macro expansion at ./REPL.jl:97 [inlined]
 [7] (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:73
while loading no file, in expression starting on line 0
WARNING: f3(x) is deprecated, use f4(x) instead.
Stacktrace:
 [1] depwarn(::String, ::Symbol) at ./deprecated.jl:70
 [2] f3(::Int64) at ./deprecated.jl:57
 [3] foo() at ./REPL[4]:3
 [4] eval(::Module, ::Any) at ./boot.jl:236
 [5] eval_user_input(::Any, ::Base.REPL.REPLBackend) at ./REPL.jl:66
 [6] macro expansion at ./REPL.jl:97 [inlined]
 [7] (::Base.REPL.##1#2{Base.REPL.REPLBackend})() at ./event.jl:73
while loading no file, in expression starting on line 0
4

julia> foo()
4

[JeffBezanson]

This seems to be because a given line number of prompt input is always considered the "same location", and warnings are uniqued by the location of the deprecated call. If you do e.g.

julia> begin
       f3(1)
       end

you get a warning because the call is now on line 2 instead of line 1.

[c42f]

Just a heads-up that I've been changing the depwarn system a bit over at #24490, which should put the deduplication logic squarely in the hands of the user, rather than being locked away in a global Set.

As part of these changes, I've set the log record id (which should be used for a fast first round of deduplication) to the raw pointer to the caller which will fix the issue you've seen here, at the cost of some duplicates for naive depwarn filtering. I anticipate having a depwarn report which can produce a legible table of all depwarns encountered while running a piece of code. Something like:

report = depwarn_report() do
    # your code here
end

The new system allows this to produce identical results each time, as the deduplication is under control of the currently installed logger.

[fredrikekre]

fixed by the logging pr