An error in the Markdown2 Header creation.

[kellertuer]

I am not sure whether this error is due to Documenter or comes from the Markdown2 utilities, but it seems that instead of just a text it expects an array of inline nodes already. How can this error appear?

Interestingly this also happens for the master branch of my project that still created a documentation successfully maybe 4 or 5 weeks ago.

ERROR: LoadError: MethodError: no method matching Documenter.Utilities.Markdown2.Heading(::Int64, ::Documenter.Utilities.Markdown2.Text)
Closest candidates are:
  Documenter.Utilities.Markdown2.Heading(::Integer, ::Array{Documenter.Utilities.Markdown2.MarkdownInlineNode,1}) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Utilities/Markdown2.jl:101
Stacktrace:
 [1] _convert_block(::Markdown.Header{1}) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Utilities/Markdown2.jl:220
 [2] iterate at ./generator.jl:47 [inlined]
 [3] collect_to!(::Array{Documenter.Utilities.Markdown2.MarkdownBlockNode,1}, ::Base.Generator{Array{Any,1},typeof(Documenter.Utilities.Markdown2._convert_block)}, ::Int64, ::Int64) at ./array.jl:651
 [4] collect_to!(::Array{Documenter.Utilities.Markdown2.CodeBlock,1}, ::Base.Generator{Array{Any,1},typeof(Documenter.Utilities.Markdown2._convert_block)}, ::Int64, ::Int64) at ./array.jl:659
 [5] collect_to_with_first!(::Array{Documenter.Utilities.Markdown2.CodeBlock,1}, ::Documenter.Utilities.Markdown2.CodeBlock, ::Base.Generator{Array{Any,1},typeof(Documenter.Utilities.Markdown2._convert_block)}, ::Int64) at ./array.jl:630
 [6] _collect(::Array{Any,1}, ::Base.Generator{Array{Any,1},typeof(Documenter.Utilities.Markdown2._convert_block)}, ::Base.EltypeUnknown, ::Base.HasShape{1}) at ./array.jl:624
 [7] map at ./array.jl:548 [inlined]
 [8] convert(::Type{Documenter.Utilities.Markdown2.MD}, ::Markdown.MD) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Utilities/Markdown2.jl:214
 [9] doctest(::Base.Docs.DocStr, ::Module, ::Documenter.Documents.Document) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/DocTests.jl:76
 [10] doctest(::Documenter.Documents.DocumentBlueprint, ::Documenter.Documents.Document) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/DocTests.jl:56
 [11] runner(::Type{Documenter.Builder.Doctest}, ::Documenter.Documents.Document) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Builder.jl:201
 [12] dispatch(::Type{Documenter.Builder.DocumentPipeline}, ::Documenter.Documents.Document) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Utilities/Selectors.jl:167
 [13] (::getfield(Documenter, Symbol("##3#8")))() at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Documenter.jl:319
 [14] cd(::getfield(Documenter, Symbol("##3#8")), ::String) at ./file.jl:96
 [15] #makedocs#1(::Bool, ::Documenter.Writers.HTMLWriter.HTML, ::Nothing, ::Nothing, ::Nothing, ::Nothing, ::Nothing, ::Nothing, ::Base.Iterators.Pairs{Symbol,Any,Tuple{Symbol,Symbol,Symbol},NamedTuple{(:modules, :sitename, :pages),Tuple{Array{Module,1},String,Array{Pair{String,Any},1}}}}, ::typeof(makedocs)) at /Users/bergmann/.julia/packages/Documenter/cMNJA/src/Documenter.jl:318
 [16] (::getfield(Documenter, Symbol("#kw##makedocs")))(::NamedTuple{(:modules, :sitename, :pages),Tuple{Array{Module,1},String,Array{Pair{String,Any},1}}}, ::typeof(makedocs)) at ./none:0
 [17] top-level scope at none:0
 [18] include at ./boot.jl:326 [inlined]
 [19] include_relative(::Module, ::String) at ./loading.jl:1038
 [20] include(::Module, ::String) at ./sysimg.jl:29
 [21] include(::String) at ./client.jl:403
 [22] top-level scope at none:0
in expression starting at /Users/kellertuer/Repositories/Software/Manopt.jl/docs/make.jl:26
```

[mortenpi]

This is the culprit: https://github.com/kellertuer/Manopt.jl/blob/b8c7bf8c69fbcd930619fcc0f572ea15475d8fb1/src/functions/differentials.jl#L71-L72

You probably want to fix that docstring and that should let your docs build with the new Documenter version. But it is also a bug in the Markdown2 conversion routines.

[kellertuer]

Nice, I would've never found that one. Yes, an empty heading is a bug – just the error message did not help to find it ;)

And yes, this is a duplicate of #1082.