Problem on opening multiple files

[melodyjulia]

The timeseries are split among two netCDF files with different number of time records. Each file can be opened with Dataset() correctly. Opening two files together with Dataset() gets wrong number of total time records.

Two netCDF files can be dragged from:
http://esgf.nci.org.au/thredds/fileServer/master/CMIP6/CMIP/CSIRO/ACCESS-ESM1-5/esm-hist/r2i1p1f1/Amon/co2/gn/v20191203/co2_Amon_ACCESS-ESM1-5_esm-hist_r2i1p1f1_gn_185001-194912.nc
http://esgf.nci.org.au/thredds/fileServer/master/CMIP6/CMIP/CSIRO/ACCESS-ESM1-5/esm-hist/r2i1p1f1/Amon/co2/gn/v20191203/co2_Amon_ACCESS-ESM1-5_esm-hist_r2i1p1f1_gn_195001-201412.nc

Use the following code to reproduce the bug:
##########################
using NCDatasets

files = ["co2_Amon_ACCESS-ESM1-5_esm-hist_r2i1p1f1_gn_185001-194912.nc","co2_Amon_ACCESS-ESM1-5_esm-hist_r2i1p1f1_gn_195001-201412.nc"]

ds = Dataset(files)

time = ds["time"][:]

close(ds)
##########################

Expected behavior

ERROR: LoadError: NetCDF error: NetCDF: Start+count exceeds dimension bound (NetCDF error code: -57)

Environment

• operating system: macOSX 10.14.6
• Julia version: 1.0.5, official binaries from https://julialang.org/downloads/]
• NCDatasets version: v0.10.3

Full output

ERROR: LoadError: NetCDF error: NetCDF: Start+count exceeds dimension bound (NetCDF error code: -57)
Stacktrace:
[1] check at /Users/test/.julia/packages/NCDatasets/NNX1o/src/errorhandling.jl:33 [inlined]
[2] nc_get_vars!(::Int32, ::Int32, ::Array{Int64,1}, ::Array{Int64,1}, ::Array{Int64,1}, ::Array{Float64,1}) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/netcdf_c.jl:916
[3] getindex(::NCDatasets.Variable{Float64,1,NCDataset}, ::StepRange{Int64,Int64}) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/variable.jl:386
[4] (::getfield(NCDatasets, Symbol("##57#58")){Tuple{StepRange{Int64,Int64}}})(::NCDatasets.Variable{Float64,1,NCDataset}) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/defer.jl:153
[5] (::getfield(NCDatasets, Symbol("##55#56")){getfield(NCDatasets, Symbol("##57#58")){Tuple{StepRange{Int64,Int64}}},NCDatasets.DeferVariable{Float64,1}})(::NCDataset{Nothing}) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/defer.jl:138
[6] #NCDataset#13(::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}, ::Type, ::getfield(NCDatasets, Symbol("##55#56")){getfield(NCDatasets, Symbol("##57#58")){Tuple{StepRange{Int64,Int64}}},NCDatasets.DeferVariable{Float64,1}}, ::String, ::Vararg{String,N} where N) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/dataset.jl:286
[7] Type at /Users/test/.julia/packages/NCDatasets/NNX1o/src/dataset.jl:284 [inlined]
[8] Type at /Users/test/.julia/packages/NCDatasets/NNX1o/src/defer.jl:137 [inlined]
[9] getindex(::NCDatasets.DeferVariable{Float64,1}, ::StepRange{Int64,Int64}) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/defer.jl:152
[10] getindex(::NCDatasets.CatArrays.CatArray{Float64,1,2,NCDatasets.DeferVariable{Float64,1}}, ::Function) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/CatArrays.jl:113
[11] getindex at /Users/test/.julia/packages/NCDatasets/NNX1o/src/multifile.jl:155 [inlined]
[12] getindex(::NCDatasets.CFVariable{Dates.DateTime,1,NCDatasets.MFVariable{Float64,1,2,NCDatasets.DeferVariable{Float64,1}},NCDatasets.MFAttributes{NCDatasets.DeferAttributes},NamedTuple{(:fillvalue, :scale_factor, :add_offset, :calendar, :time_origin, :time_factor),Tuple{Nothing,Nothing,Nothing,String,DateTimeProlepticGregorian,Int64}}}, ::Colon) at /Users/test/.julia/packages/NCDatasets/NNX1o/src/cfvariable.jl:518
[13] top-level scope at none:0
[14] include at ./boot.jl:317 [inlined]
[15] include_relative(::Module, ::String) at ./loading.jl:1044
[16] include(::Module, ::String) at ./sysimg.jl:29
[17] exec_options(::Base.JLOptions) at ./client.jl:266
[18] _start() at ./client.jl:425

[Alexander-Barth]

Thank you for the bug report. I could reproduce the issue and committed a fix in the master branch. I will make a new release shortly.

Fell free to reopen the issue if the fix does not resolve your problem.

[melodyjulia]

Thank you very much!