# Broken Base.show method

**URL:** <https://discourse.julialang.org/t/broken-base-show-method/48453>\
**Category:** General Usage\
**Created:** [October 15, 2020, 7:20pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453 "2020-10-15T19:20:38Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![racinmat](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/racinmat/32/11715_2.png) [@racinmat](https://discourse.julialang.org/u/racinmat)\
**Post date:** [October 15, 2020, 7:20pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/1 "2020-10-15T19:20:38Z")

</div>

I’m working on function that either shows full types or not based on some variable.  
I encountered weird bug I can’t figure out because the function is not behaving the same when running it and when debugging it.  
The function is here: [https://github.com/racinmat/Mill.jl/blob/master/src/Mill.jl#L74](https://github.com/racinmat/Mill.jl/blob/master/src/Mill.jl#L74)

When I run it, it crashes with

```julia
ERROR: type DataType has no field var
Stacktrace:
 [1] getproperty at .\Base.jl:28 [inlined]
 [2] show(::IOContext{Base.GenericIOBuffer{Array{UInt8,1}}}, ::Type{ERROR: type DataType has no field var

```

but when I use Debugger and debug it line per line, Julia executes the code succesfully, it does not crash, it returns correctly.  
During debugging, `x isa DataType` is true, but when running it outside of debug, it’s false.  
I restarted Julia multiple times to make sure it’s not some old version of code in the REPL.

The function itself is here:

```julia
function Base.show(io::IO, x::Type{T}) where {T<:Union{AbstractNode,AbstractMillModel}}
    if _terseprint[]
        if hasproperty(x, :body) && !hasproperty(x.body, :name) && hasproperty(x.body, :body)
            print(io, "$(x.body.body.name){…}")
            return
        elseif hasproperty(x, :body) && !hasproperty(x, :name)
            print(io, "$(x.body.name){…}")
            return
        else
            print(io, "$(x.name){…}")
            return
        end
    else
        # specifically function show(io::IO, @nospecialize(x::Type))
        if x isa DataType
            Base.show_datatype(io, x)
            return
        elseif x isa Union
            if x.a isa DataType && Core.Compiler.typename(x.a) === Core.Compiler.typename(DenseArray)
                T2, N = x.a.parameters
                if x == StridedArray{T2,N}
                    print(io, "StridedArray")
                    Base.show_delim_array(io, (T2,N), '{', ',', '}', false)
                    return
                elseif x == StridedVecOrMat{T2}
                    print(io, "StridedVecOrMat")
                    Base.show_delim_array(io, (T2,), '{', ',', '}', false)
                    return
                elseif StridedArray{T2,N} <: x
                    print(io, "Union")
                    Base.show_delim_array(io, vcat(StridedArray{T2,N}, Base.uniontypes(Core.Compiler.typesubtract(x, StridedArray{T2,N}))), '{', ',', '}', false)
                    return
                end
            end
            print(io, "Union")
            Base.show_delim_array(io, Base.uniontypes(x), '{', ',', '}', false)
            return
        end

        if Base.print_without_params(x)
            return show(io, Base.unwrap_unionall(x).name)
        end

        if x.var.name === :_ || Base.io_has_tvar_name(io, x.var.name, x)
            counter = 1
            while true
                newname = Symbol(x.var.name, counter)
                if !Base.io_has_tvar_name(io, newname, x)
                    newtv = TypeVar(newname, x.var.lb, x.var.ub)
                    x = UnionAll(newtv, x{newtv})
                    break
                end
                counter += 1
            end
        end

        show(IOContext(io, :unionall_env => x.var), x.body)
        print(io, " where ")
        show(io, x.var)
    end
end

```

I’m desperate, I can’t figure out how to make it work and make it consistent.

I know one should generally not overload `Base.show`, but we have good usecase for this.

The error can be seen here: [Travis CI - Test and Deploy with Confidence](https://travis-ci.com/github/racinmat/Mill.jl/jobs/400338894)

---

<div class="post-metadata">

**Author:** ![kristoffer.carlsson](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/kristoffer.carlsson/32/22_2.png) [@kristoffer.carlsson](https://discourse.julialang.org/u/kristoffer.carlsson)\
**Post date:** [October 15, 2020, 7:28pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/2 "2020-10-15T19:28:07Z")

</div>

Often when it works in the debugger and not the compiler, it’s a bug in the compiler. The best thing to do is probably to try reduce the example as much as possible and open an issue.

---

<div class="post-metadata">

**Author:** ![tomerarnon](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tomerarnon/32/3170_2.png) [@tomerarnon](https://discourse.julialang.org/u/tomerarnon)\
**Post date:** [October 15, 2020, 7:58pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/3 "2020-10-15T19:58:04Z")

</div>

See also [https://github.com/JuliaLang/julia/pull/37773#issuecomment-701976134](https://github.com/JuliaLang/julia/pull/37773#issuecomment-701976134)  
and the discussion in [These stack traces are out of control - #6 by tomerarnon](https://discourse.julialang.org/t/these-stack-traces-are-out-of-control/47162/6)  
This sounds related.

---

<div class="post-metadata">

**Author:** ![racinmat](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/racinmat/32/11715_2.png) [@racinmat](https://discourse.julialang.org/u/racinmat)\
**Post date:** [October 15, 2020, 8:16pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/4 "2020-10-15T20:16:35Z")

</div>

Oh, thanks. I’ll try to make an MWE and open issue if I’ll be successful.  
Is there some workaround what could I do so it would be working for me?  
Make different function for `_terseprint[] == true` and another for `_terseprint[] == false` hoping that function barrier could avoid this bug?

---

<div class="post-metadata">

**Author:** ![simeonschaub](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/simeonschaub/32/216566_2.png) [@simeonschaub](https://discourse.julialang.org/u/simeonschaub)\
**Post date:** [October 15, 2020, 8:20pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/5 "2020-10-15T20:20:17Z")

</div>

I would say that overloading printing for union types is definitely piracy, so it should generally be avoided. That said, it does indeed look like there’s something else going wrong here.

---

<div class="post-metadata">

**Author:** ![Sukera](https://avatars.discourse-cdn.com/v4/letter/s/ce7236/32.png) [@Sukera](https://discourse.julialang.org/u/Sukera)\
**Post date:** [October 15, 2020, 8:24pm UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/6 "2020-10-15T20:24:08Z")

</div>

If it’s really reproducible, you could create an rr-trace and submit that.

> **[Julia 1.5 Feature Preview: Time Traveling (Linux) Bug Reporting](https://julialang.org/blog/2020/05/rr/)**
>
> Julia 1.5 is gaining a cool new bug reporting capability, leveraging mozilla's rr project to automatically create fully-reproducible bug reports

---

<div class="post-metadata">

**Author:** ![marius311](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/marius311/32/3953_2.png) [@marius311](https://discourse.julialang.org/u/marius311)\
**Post date:** [October 16, 2020, 12:15am UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/7 "2020-10-16T00:15:32Z")

</div>

I’ve run into some weird stuff with overloaded `show` methods, the root cause of which is this [https://discourse.julialang.org/t/dispatching-on-the-result-of-unwrap-unionall-seems-weird](https://discourse.julialang.org/t/dispatching-on-the-result-of-unwrap-unionall-seems-weird)

I’m not sure its related to your problem (in particular not sure why debug mode would matter), but there certainly are alot of the same ingredients as in your code snippet so figured I’d mention.

---

<div class="post-metadata">

**Author:** ![tim.holy](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tim.holy/32/52_2.png) [@tim.holy](https://discourse.julialang.org/u/tim.holy)\
**Post date:** [October 16, 2020, 11:18am UTC](https://discourse.julialang.org/t/broken-base-show-method/48453/8 "2020-10-16T11:18:13Z")

</div>

I am paranoid about overloading `show` for types. It often makes it hard to debug problems, and there’s a long history of weird behavior associated with such overloads. The best usage of overloading `show` for types I’ve ever seen was for Unitful.jl, and even they decided to stop doing it: [https://github.com/PainterQubits/Unitful.jl/issues/321](https://github.com/PainterQubits/Unitful.jl/issues/321).