Re: question regarding `c' operator acting on rectangular region

[joerg van den hoff]

On 27.05.23 13:27, Paul Fox wrote:
> joerg van den hoff wrote:
>   > rarely using rectangular regions (and when, then for deletion), so
>   > I was surprised by the following:
>   >
>   > issuing something like
>   >
>   > c^Aqjq
>   >
>   > to select a rectangle with a width of one character and a height of
>   > 2 lines and trying to replace the rectangle by a multicharacter
>   > string fails.  only the first character of the replacement string
>   > specified in response to the "Rectangular text:" query is inserted.
>   >
>   > e.g. with an input of
>   >
>   > 1 2 3
>   > 1 2 3
>   >
>   > and selecting the 2nd column as rectangle and trying to replace the
>   > 2s by "abc" results in
>   >
>   > 1 a 3
>   > 1 a 3
>   >
>   > I would have expected the text to the right of the rectangle to be
>   > shifted to the right to make room for the full replacement string?
>   >
>   > questions:
>   >
>   > * am I doing it wrong?
>
> I'm afraid not.  This is the current behavior.

understood. thank you for clarifying.

>   >
>   > * if no, is this to be expected behaviour or a bug?
>   >
>   > * if not a bug, is there a canonical way to achieve the desired
>   > * edit result, i.e.
>   >
>   > 1 abc 3
>   > 1 abc 3
>
> No easy way.  I find myself using ^A-r to open a rectangle with the
> right amount of space (and it's usually too much :-/ ), then put in
> the text I want, then delete the extra, if any.  I don't know how hard
> it would be internally to force the new empty space in order to make
> room for the new string.  Currently if the new text is shorter than
> the rectangle width, it's padded with spaces -- to be consistent, the
> rectangle should probably be shrunk to match the string length.

yes, I see. I was considering that as a work around if need be. but then
settled on (m)awk an hour ago :).

> I seem to recall I wasn't terribly happy with the way the rectangular
> operations turned out, after implementing them, and I think this was
> part of it.

they are still quite useful without doubt...

but regarding the `c' operator it would sure be nicer if it would behave on rectangles like it does 
on, say, words: adjust the selected region's width to the replacement string width. since ^A-r is
available/working (and the underlying code could be used to inject something else than blanks I 
guess), I wonder how difficult it would be to make c^A-q do the "expected" thing?

but in any case thanks again for explaining the situation. I just didn't know. and one of course can 
work around the given limitation.

joerg

> paul
>
>   >
>   > in this example?
>   >
>   > thank you
>   > joerg
>   >
>
>
> =----------------------
> paul fox, pgf@foxharp.boston.ma.us (arlington, ma)