Output delimiter should be same as input

[mbhall88]

cut automatically uses the input delimiter as the output delimiter. I find it quite annoying to have to specify the delimiter twice. I wonder if you would consider changing the default behaviour to do so? I appreciate this is a breaking change, but might be better to do this before hitting v1?

[sstadick]

Hi! I would lean toward no because a regex can be used as the delimiter and the behavior in that scenario could be surprising. But I'll think about it a bit more and leave this issue open for now.

The other bit of my reasoning is that while iterating on a one-liner, tabs are preferable to look at the outputs, then, once you've settled on the command, it's not too onerous to switch to a different output delimiter.

[sstadick]

@ghuls, as the only other active user of hck that I know of, do you have any thoughts on this?

[ghuls]

@sstadick I generally work with TAB delimited files and not often with files with other delimiters. I use it for processing of multi gigabyte files and not for small text files, so typing a few extra characters does not bother me. TABs by default make everything more readable imho. awk also uses \s+ as field separator and as output separator.

The need to add a dolar sign before '\t and \n, when used as output separator, annoys me more. It would be nice if those are recognized by default.

❯  hck -d ':' -D $'\t' /etc/group | head
root	x	0	
bin	x	1	
daemon	x	2	
sys	x	3	
adm	x	4	
tty	x	5	
disk	x	6	
lp	x	7	
mem	x	8	
kmem	x	9	

❯  hck -d ':' -D '\t' /etc/group | head
root\tx\t0\t
bin\tx\t1\t
daemon\tx\t2\t
sys\tx\t3\t
adm\tx\t4\t
tty\tx\t5\t
disk\tx\t6\t
lp\tx\t7\t
mem\tx\t8\t
kmem\tx\t9\t

❯  hck -d ':' -D '\n' /etc/group | head
root\nx\n0\n
bin\nx\n1\n
daemon\nx\n2\n
sys\nx\n3\n
adm\nx\n4\n
tty\nx\n5\n
disk\nx\n6\n
lp\nx\n7\n
mem\nx\n8\n
kmem\nx\n9\n

/v/leuven-user/303/vsc30366 vsc30366 in r23i27n23 in /v/leuven-user/303/vsc30366 via 🐏  467GiB/756GiB 62% 
❯  hck -d ':' -D $'\n' /etc/group | head
root
x
0

bin
x
1

daemon
x

[sstadick]

That's a good point, the $ is pretty annoying.

[sstadick]

#47 will fix the unescaping for -d and -D.

@mbhall88 I've compromised and added a -I options that will reuse the input delimiter as the output delimiter if the intput delimiter is a literal and no other output delim is specified. So to replicate cut you could now do something like: hck -LId' ' <file> and the space will be reused for the output.

[ghuls]

@sstadick Thanks for fixing the unescaping. Works much more convenient now.

[sstadick]

@ghuls see #49 / v0.7.1 for incoming fix for the header line parsing to apply the same unescaping.

[sstadick]

Closing for now, please reopen if needed.