Re: "drop-in replacement" and transpose-words-match

[Daniel Shahaf <d.s@xxxxxxxxxxxxxxxxxx>]

[ Sorry for the late reply, I missed this mail last week. ]

Bart Schaefer wrote on Mon, Jan 11, 2016 at 23:32:59 -0800:
> Of the eight match-words-by-style functions, transpose-words-match is the
> only one that is not really a drop-in, because it adopts the paradigm of
> match-words-by-style where a cursor in the middle of a word splits that
> word (treating the LBUFFER part as "word before the cursor" and RBUFFER
> part as "word after the cursor").
> 
> There are some useful things about this behavior, e.g., "skip-chars 1" 
> where you can make "fooXbar" swap as "barXfoo", but there are also some
> drawbacks -- cases where transpose-words would do something but -match
> does nothing or does something very different.  So how about this:  if
> the skip-chars style is set (even to zero), then keep the cursor where
> it is and split the word, otherwise act like the builtin.  The subword
> flavor of word-style can similarly be checked.
> 

That wouldn't let a user have the skip_chars=0 beahaviour for a wider
context and the "act like the builtin" behaviour for a more specific
subcontext.  (It's not possible to unset for a subcontext a style that's
set on a wider context.)

How about changing the pattern from «*» to «<->»?  Then a setting on
a wider context can be hidden for a subcontext by setting the style to
'' on the subcontext.

Cheers,

Daniel


> +++ b/Functions/Zle/transpose-words-match
> @@ -11,14 +11,23 @@
> +if [[ $WIDGET == transpose-words ]]; then
> +  # default is to be a drop-in replacement, check styles for change
> +  zstyle -m $curcontext skip-chars \* ||
> +  zstyle -m $curcontext word-style '*subword*' ||
> +  { [[ $LBUFFER[-1] != [[:space:]] && $RBUFFER[1] != [[:space:]] ||
> +       -z ${RBUFFER//[[:space:]]/} ]] && zle backward-word }
> +fi