loadtable with missing values given by an empty cell

[MaximilianJHuber]

Hi,

I read couple of .csv files with seperator ';' and 36 columns. I specify the column types correctly. The first two lines are read:

100;20030701;20150831;1234567890;20070131; <...>
200;19991018;20150831;1324567432;20070131; <...>

The third line not:

300;20070101;20070624;;20070131;

although I set "" as nastring:

loadtable(path * file * "/", output="bin", chunks=8, delim=';', distributed=false, header_exists=false, colnames = names, colparsers = [Int64, String, String, Int64, <...>], nastrings=[""])

The error message is:
MethodError: Cannot convert an object of type DataValues.DataValue{Int64} to an object of type TextParse.StrRange
and
CSV parsing error in C:\data.csv at line 4 char 22: ...300;20070101;20070624;;20070131; <...>

One bug is, that with head_exists=false the line number of the error is wrong, it is the thrid line, not the fourth.
And my main question is, how do I tell loadtalbe to take and empty cell as Missing?

[MaximilianJHuber]

Not specifying the colparsers and nastrings, but increasing the type_detect_rows was successful. I now understand that a possibly missing integer column needs to be specified as DataValues.DataValue{Int64}.

[shashi]

So what's the issue here now? would you say it's lack of documentation for this case?

[MaximilianJHuber]

Yes, lack of documentation for handling of missing values, and if header_exists=false the line reported responsible for an error should be reduced by one.