InexactError in join

[ExpandingMan]

I am experiencing an extremely difficult-to-reproduce InexactError when using join in master.

ERROR: InexactError()
 in convert(::Type{UInt32}, ::Int64) at ./int.jl:205
 in sharepools(::DataFrames.DataFrame, ::DataFrames.DataFrame) at /home/savastio/.julia/v0.5/DataFrames/src/abstractdataframe/join.jl:204
 in #join#46(::Array{Symbol,1}, ::Symbol, ::Function, ::DataFrames.DataFrame, ::DataFrames.DataFrame) at /home/savastio/.julia/v0.5/DataFrames/src/abstractdataframe/join.jl:283
 in (::Base.#kw##join)(::Array{Any,1}, ::Base.#join, ::DataFrames.DataFrame, ::DataFrames.DataFrame) at ./<missing>:0

The join is being done on two DateTime and one Int64 columns. I attempted to reproduce the error by creating test dataframes with the same types and performing a similar join, but this does not seem to produce any error. The error also goes away if I join only on one of the DateTime columns and the Int64 column. At the moment I'm not able to determine what is special about the dataframes that are causing the crash.

Wish I could be more helpful, but I thought I'd post the issue in case it was obvious to someone what was going on.

[nalimilan]

I guess the two columns have many unique values? Then that would be a duplicate of #985 (since the code is the same as for groupby).

[ExpandingMan]

Yes, the columns have many unique values, so that is likely it. Thanks.

[nalimilan]

Closing then. We really need somebody to investigate how we can avoid this failure.

[diegozea]

Hi! I found the same error:

julia> const pdb_evalue = join(pdb_selected, selected, on = [:pfam, :seq])
ERROR: InexactError()
Stacktrace:
 [1] convert(::Type{UInt32}, ::Int64) at ./int.jl:332
 [2] PooledDataVecs(::DataFrames.DataFrame, ::DataFrames.DataFrame) at /home/diego/.julia/v0.6/DataFrames/src/abstractdataframe/join.jl:93
 [3] #join#52(::Array{Symbol,1}, ::Symbol, ::Function, ::DataFrames.DataFrame, ::DataFrames.DataFrame) at /home/diego/.julia/v0.6/DataFrames/src/abstractdataframe/join.jl:202
 [4] (::Base.#kw##join)(::Array{Any,1}, ::Base.#join, ::DataFrames.DataFrame, ::DataFrames.DataFrame) at ./<missing>:0

Is there a workaround to avoid it?

Best regards,

Julia 0.6.1
DataFrames 0.10.1

[nalimilan]

Is there a workaround to avoid it?

Use DataFrames master... :-)

[diegozea]

Sadly, Pkg.checkout("DataFrames") gives me the following error:

Unsatisfiable requirements detected for package WeakRefStrings:
├─version range [0.4.0,∞) required by package DataFrames, whose only allowed version is 0.10.1+:
│ └─version 0.10.1+ set by fixed requirement (package is checked out, dirty or pinned)
└─version range [0.0.0,0.3.0) required by package CSV, whose allowed version range is [0.0.0-,∞):
  └─version range [0.0.0-,∞) set by an explicit requirement
The intersection of the requirements is empty.