header option eats extra line

[duffee]

Using mtcars_001.csv file from csvkit_talk (a file that has one header line and 32 rows of data) and the perldl shell, I get

pdl> $cars = rcsv2D('mtcars_001.csv', {text2bad => 1, header => 0, debug => 1})                                  
Fetching 2D chunk=40000, reshape=80000, bad=?/1, sep_char=','
Initialized size=80000, cols=12, type=double
Reshaping to 33 (final)

and with the header option turned on, I get

pdl> $cars = rcsv2D('mtcars_001.csv', {text2bad => 1, header => 1, debug => 1})                                  
Fetching 2D chunk=40000, reshape=80000, bad=?/1, sep_char=','
Initialized size=80000, cols=12, type=double
Reshaping to 31 (final)

which skips the first row of data. The same thing happens for higher values in that one more line than specified is skipped. Setting the value of header to either -1 or 0.5 fixes the problem

pdl> $cars = rcsv2D('mtcars_001.csv', {text2bad => 1, header => -1, debug => 1})                                 
Fetching 2D chunk=40000, reshape=80000, bad=?/1, sep_char=','
Initialized size=80000, cols=12, type=double
Reshaping to 32 (final)

version info:

• PDL::IO::CSV v0.010
• PDL v2.019 (supports bad values)
• perlDL shell v1.357
• perl v5.26.1