readxlsheet reads the empty cells as #NA

[ezgikurt]

Hi, I'm using Julia 0.6.4, with version 0.9.0. This used to work fine in the earlier versions but I've recently updated the package to 0.9 and now readxlsheet reads the empty cells in the file as "#NA" and julia does not register this as a missing value. typeof() returns DataValue(Union{})().

I need these to register as missing values, are there any suggestions?

[davidanthoff]

I don't think anything changed there? What behavior would you expect?

[ezgikurt]

They used to be registered as "missing", and functions such as ismissing() returned true (now they return false).

[davidanthoff]

Ah, I think you were reading them into a DataFrame, right? That functionality has been moved from this package to ExcelFiles.jl, and if you materialize things into a DataFrame that way, you'll get missing values for empty cells.

I don't think this package here ever created missing values, though...

[ezgikurt]

That's the case with ExcelFiles,but I usually used ExcelReaders like so and it did create missings (NAs in earlier versions of DataArrays).

a = readxlsheet(filename, sheetname);

a would come back as a DataArray{Any,2} with missings, and we did not pre-allocate it.

[davidanthoff]

Ah, interesting! I guess what happened is that DataArray was ported to Missing at some point, and then all of a sudden ExcelReaders.jl returned missing values, even though I was not even aware of that :)

The Queryverse is not using the whole Missing story. That design doesn't work for some key scenarios due to some technical limitations, so Queryverse is using DataValues.jl for its missing value story, and that is what you are getting there.

[ezgikurt]

Oh all right, thanks for the responses! Going forward I'm going to have to use ExcelFiles I guess.