Re: CSV extension status

[Neil R. Ormos]

Manuel Collado wrote:
> Neil R. Ormos escribio:

>> What would be really useful to me would be a
>> facility that transparently and efficiently
>> treats real-world CSV input as though the
>> equivalent input had been received as
>> FS-separated input, with sensible handling of
>> embedded newlines.  By "transparent" I mean
>> with the same side-effects and without
>> additional vexatious restrictions.

> That is exactly the goal of gawk-csv with
> CSVMODE=1, and CSVMODE with CSVMODE=-1. The
> original CSV notation is discarded and clean
> text field values are delivered, separated by a
> suitable OFS mark (null char, or SUBSEP, to
> prevent interference with real-world data).

Do either the gawk-csv extension or the CSVMODE library provide fully 
transparent handling of input records?

Are all of the usual public global variables that are set when a regular record 
is read set in a parallel fashion, including RT and all the rest?

Are assignments to, for example, $5 transcribed to $0 in exactly the same way 
they would be in the regular case?

Can fields contain NUL=$00 characters?

(I'm just trying to understand how the extension and library work, not 
criticising design decisions, which I assume were made for good reasons.)