Short option for --nidx --fs tab

[sjackman]

The TSV file format of cut and friends is sufficiently common that I feel that --nidx --fs tab warrants a short option, similar to -p, perhaps -t. It's particularly handy when using mlr to replace commands from coreutils, for example, mlr -t cut -f 3. Your thoughts?

I'd also suggest a short option for --tsvlite, perhaps -T.

[johnkerl]

Cool; mlr put already uses -t and -T but mlr main uses neither.

[sjackman]

Awesome. Do you have a preference which of the two is -t and which is -T? Which format do you think is more common? I'd say that the more common one should be -t.

[johnkerl]

@sjackman i don't have a preference. --tsvlite = -T is fine with me.

[sjackman]

I use --tsvlite more than I use --nidx --fs tab. I'm thinking
-t is --tsvlite and
-T is --nidx --fs tab

[sjackman]

Woohoo! Thanks, John. New release pending?

[johnkerl]

Pretty soon ... I want to get the comment-skipper in there too and it needs a bunch of regression tests, and one more corner-case to implement.

[sjackman]

Awesome. While you're doing the comment-skipper, I have a use case to pass all comments through unmodified to the output. Think that could make it in?

[johnkerl]

bler, i don't know how to do that within the miller streaming architecture ...

[sjackman]

I was thinking as it's read in from the input, if it's a comment, immediately write that same line out to the output.

[johnkerl]

gler, that would be ok with line-at-a-time mappers but w sort etc it's not clear to me what would be the right thing to do.

[sjackman]

Print the comment lines out in the order that they're observed. In the case of sort, the comment lines would appear before the normal records in the order that they're observed, which would be consistent with a stable sort where a comment line sorts before any other record. That happens to be the behaviour that I need. In my case, the comment lines all appear at the top of the file before the normal records.