Re: [9front] [PATCH] rio: add -label wctl param; improve window(1)

[kvik@a-b.xyz]

> Good in principle, I think, but just to note: if window(1) no longer calls
> itself, users will probably have to change quoting in scripts which call
> window.

I may be missing something but I don't see how this would be the case.

The arguments constituting a command are evaluated once by the calling
shell and are passed into window(1), ending up in the $cmd list.  Before
or after this patch this list stays the same as it is passed along;
specifically it does not get reevaluated until the final rc -c in either
rio itself or in (new) window(1).  Consider the example:

	fn nargs {
		echo $#* : $*
	}
	fn once {
		nargs $*
		twice $*
	}
	fn twice {
		nargs $*
		echo 'nargs '$"*
		rc -c 'nargs '$"*
	}

	; once 1 '2 3' '''4 5'''
	3 : 1 2 3 '4 5'
	3 : 1 2 3 '4 5'
	nargs 1 2 3 '4 5'
	4 : 1 2 3 4 5

You see that the initial argument list stays the same as it is being
passed along, that is until another round of evaluation inside the
rc -c subshell where the next layer of quotes is stripped.

In the argument parsing sense removing the recursive call to window -x
is the same as skipping the call to 'once' in the above example.

> window -r 268 0 868 400 -scroll rc -c \
> 	'''label -f kprint; echo kprint; tail -f /tmp/kprint'''
> window -r 0 0 128 128 colorclock_l
> window -r 0 124 128 252 rc -c \
> 	'''font=/lib/font/bit/dejavusans/unicode.12.font; stats -Ime'''

These commands run as expected with the patch.

> Now I think of it, I don't really agree with window(1) setting the label
> at all, but it's a minor point.

I'm not a fan of automatic labeling either.  Ideally the programs would
set the label themselves if the user didn't request a name, but I don't
think this is easily accomplished.