CSV.parse(csv, col_sep: nil) throw a TypeError (no implicit conversion of nil into String) in ruby 2.6 about csv HOT 8 CLOSED

OK.
We don't accept nil for col_sep:.

from csv.

It's not intended but could you show any useful use case for col_sep: nil?

from csv.

I don't have a particularly good example. But that the lib "crash" does not seem to me desirable. Go back to the previous behavior, or handle the col_sep: nil with an ArgumentError seems more better to me.

from csv.

I have a use case for col_sep: nil, but I don't know if it is sound. I want to parse different sources, csv and flat files. I'm showing you a couple of examples. A csv file:

field1, field2
abc, def

and a flat file.

ghi
jk,l

The flat file includes the comma. To handle those 2 files (think not 2 files but tens of files csv and flat), I'd like to use the csv parser, passing col_sep: nil when reading the flat file. In this case nil would mean "I don't want a column separator". Is this a valid use case in your opinion?

from csv.

I think that File.foreach("flat.txt", chomp: true) {|line| ...} should be used for the flat file.
We should use the csv library only for a CSV (family) file. :-)

from csv.

Yes, there are a bunch of methods to process a text file. However, a flat file can be considered a CSV with a single column and without an header. If you code is fed by both "real" csv and "semi" csv (flat files), the idea to have a single piece of code that handle them all could bo profitable.

from csv.

I don't want to recommend the "flat file is a CSV file" approach...
It will confuse many programmers...

from csv.

I know, that was just to share a possible approach and to share some feedback.

from csv.