experimental, add total match count

[bounceme]

if @romainl is into it. this will probably need to be optional

[romainl]

Thanks, I'll wait until this week-end to try it locally.

[auwsmit]

Great patch. If romainl doesn't like it, it'd be great as its own separate plugin.

And a suggestion: putting the search term after the numbers. Like:

/match 123 of 456 for searchterm

As it is now going before, the match results will have an inconsistent position depending on the length of your query.

Alternatively, just don't show the search term at all, since it's already shown by its matching results. Like:

match 123 of 456

It's the smallest of gripes though and I can understand wanting to keep the search term in its same place for pure consistency.

[auwsmit]

Just found a bug: Exits visual mode when cursor goes over a search result with hlsearch on. Probably from the norm!

[bounceme]

i'll look for the cause of that bug, thanks. my main focus is on making this as close to the default behaviour as possible, also if the timeout is reached it doesn't break your concentration by looking really different

[auwsmit]

Found another bug on my end: any time a search mapping (e.g. *, n, etc.) has zz at the end, it strangely causes the results message to disappear most of the time.

To reproduce: map n nzz and compare n to N. The message should usually disappear for n when zz moves the screen, but never for N

edit: could just be gvim on Windows, I haven't tested in Unix gvim or a terminal because Windows blows

[bounceme]

							*z.*
z.			Redraw, line [count] at center of window (default
			cursor line).  Put cursor at first non-blank in the
			line.

							*zz*
zz			Like "z.", but leave the cursor in the same column.
			Careful: If caps-lock is on, this command becomes
			"ZZ": write buffer and exit!

redraw.

[auwsmit]

Ah, I didn't catch that, thanks. Also, random suggestion that might be dumb or pointless:

Could this

exe "silent! norm! :let g:cool_char=nr2char(screenchar(screenrow(),1))\<cr>"
if g:cool_char =~ '[/?]'
  ...
endif

be replaced with just

let g:cool_char = v:searchforward ? '/' : '?'
...

?

[bounceme]

no, v:searchforward reflects whether you typed the pattern after pressing ? or /

"this seems to work
augroup PCOOL
  au!
augroup END
nnoremap <silent>n :<C-U>set scrolloff=999\|au PCOOL cursormoved * set scrolloff&vim\|
      \ au! PCOOL cursormoved *<cr>:call feedkeys(max([v:prevcount,1]).'n','n')<cr>

[auwsmit]

Sorry if I'm misunderstanding, but v:searchforward in fact reflects the last direction of the search. Based on my testing, it's functionally equivalent to check that variable rather than the lowerleft-most character of the window (as I showed above), and seems like a more optimal approach.

Also good idea with scrolloff to center without redrawing

[bounceme]

doesn't reflect the movement direction, like vim by default does

[auwsmit]

What's an example of g:cool_char = v:searchforward ? '/' : '?' giving the wrong character?

[bounceme]

we don't know which direction the last movement was in.

?let<cr>N echoes /let by default

[auwsmit]

Oh... I was using my vimrc where I have the results bit put into a function that's attached to the end of all my search mappings, so v:forwardsearch was always up-to-date and worked flawlessly for me. But you're right that using the fancy vim-cool method, it's totally wrong. Sorry for wasting your time, I was up very late last night.

[bounceme]

guess #23 also should apply to the counting

[romainl]

@bounceme, this looks good and useful but I'd prefer to have it behind an option as it is not directly related to the original purpose of vim-cool.

[bounceme]

merging it. thanks for the Readme updates @romainl