I'm planning to implement a csv loader, but I have some questions that I can't answer before implement it.
1. Should I decide on dtype right in C++ code or just parse csv as string arrays then cast them in Ruby?
2. Can I assume that csv file is 'comma separated' ? I mean, in some case, csv files are separated by tabs or something else.
If the answer to the first question is 'no', the situation will be quite simple.
If the answers to the questions above are both 'yes', I may write up some flex codes(a parser generator) to parse the csv file.
But If the second is 'no', while the first is 'yes', things may be a little complicated, that I may have to write a DFA by myself.