New test failing ("ml-10m")
Closed this issue · 3 comments
I'm getting a really strange error in a local test that I can't seem to figure out why.
I've attached a screenshot, but it says that there's missing "end" with the "begin" at line 115 of the graphs_no_ci.jl test file, but the "end" seems to be there.
I'd like to know if I am the only one experiencing this.
It happens on Julia 1.7.3 running Ubuntu 22.04.
I also tested it on Windows 11 and the same error appeared.
Let me look into this.
Hi @Dsantra92, the patch fixed the original error, but now I'm getting a new error.
Testing Running tests...
[ Info: Testing larger datasets
graphs_no_ci.jl: Error During Test at /home/christian/.julia/packages/MLDatasets/fdW2b/test/runtests.jl:48
Got exception outside of a @test
LoadError: syntax: incomplete: "begin" at /home/christian/.julia/packages/MLDatasets/fdW2b/test/datasets/graphs_no_ci.jl:115 requires end
Stacktrace:
[1] top-level scope
@ ~/.julia/packages/MLDatasets/fdW2b/test/datasets/graphs_no_ci.jl:115
[2] include(fname::String)
@ Base.MainInclude ./client.jl:451
[3] macro expansion
@ ~/.julia/packages/MLDatasets/fdW2b/test/runtests.jl:49 [inlined]
[4] macro expansion
@ ~/.julia/juliaup/julia-1.7.3+0~x64/share/julia/stdlib/v1.7/Test/src/Test.jl:1359 [inlined]
[5] macro expansion
@ ~/.julia/packages/MLDatasets/fdW2b/test/runtests.jl:48 [inlined]
[6] macro expansion
@ ~/.julia/juliaup/julia-1.7.3+0~x64/share/julia/stdlib/v1.7/Test/src/Test.jl:1283 [inlined]
[7] top-level scope
@ ~/.julia/packages/MLDatasets/fdW2b/test/runtests.jl:42
[8] include(fname::String)
@ Base.MainInclude ./client.jl:451
[9] top-level scope
@ none:6
[10] eval
@ ./boot.jl:373 [inlined]
[11] exec_options(opts::Base.JLOptions)
@ Base ./client.jl:268
[12] _start()
@ Base ./client.jl:495
in expression starting at /home/christian/.julia/packages/MLDatasets/fdW2b/test/datasets/graphs_no_ci.jl:115
I managed to get rid of the error by changing the '!' in line 365 of "src/datasets/graphs/movielens.jl" to a ':', but this feels like a hack.
I did a bit of digging and it does not happen on Windows. The output of typeof(tag_df[:,3]
on Windows and Linux is Vector{String} (alias for Array{String, 1})
, while the output of typeof(tag_df[!,3]
on Windows is the same, but on Linux, it's SentinelArrays.ChainedVector{String, Vector{String}}
, which is what seems to be causing the error.
Upon further examination, it seems like the text is not being read in properly for 5 identical lines. the lines are lines 50208, 50572, 50593, 52325, 52891 of the ml-10m "tags.dat". I believe the presence of "=
in the string is causing CSV to assign undef
.
This happens with a !
or a :
, so fixing with a colon would have only hidden the issue.