inconsistency of -d

[ggrothendieck]

For enum -d refers to the input and output:

>xan range 3 | xan map "n as n2" | xan fmt -t? | xan -d? enum
index?n?n2
0?0?0
1?1?1
2?2?2

whereas on transform it refers to the input only:

> xan range 3 | xan map "n as n2" | xan fmt -t ? | xan enum -d? | xan transform -d? * "_**2"
index,n,n2
0,0,0
1,1,1
4,4,4

In Miller (mlr) one can specify both the input field separator (--ifs) and the output delimiter (--ofs) on most commands. In qsv -d is the input delimiter and -t is the output delimiter. awk has -F as the input field separator and -v OFS= as the output field separator.

[Yomguithereal]

-d is always consistent and refers to the input delimiter indeed. Now some commands are not consistent with the general logic of the command which is to always output "standard" CSV (commas, double quotes etc.) for performance purposes (this typically includes xan behead & xan rename, where no normalization of the input is performed so the stream can be forwarded as-is). Now I agree xan enum is not particularly explicit about this and its help could be improved.

Now there could be a flag for output delimiter but it would be quite a hassle to add it to all commands and you can always have a xan fmt pipe at the end of your pipe if required. I am not opposed to adding an environment variable for this though.

[CodeMan99]

xan rename, where no normalization of the input is performed so the stream can be forwarded as-is).

I am not following. Should I have a file that is tab-separated, then pipe that into this command, I'll end up with a comma-separated header on top of my data that is still tab separated.

$ printf "%s\t%s\n" Name Age \
>                   Bob    30 \
>                   Jeff   32 > example.tsv
$ xan rename -d $'\t' 'First Name,Age' example.tsv
First Name,Age
Bob     30
Jeff    32

This output seems to be invalid. In fact, xan view fails regardless of what delimiter is given.

[Yomguithereal]

@CodeMan99 this looks like a xan rename bug. Let me take care of that.

[Yomguithereal]

@CodeMan99 a fix was introduced in last commit.