Better isearch integration about consult HOT 27 CLOSED

There is the better solution - orderless-transient-matching-styles

Oh, that's right! I completely forgot. Thank goodness there is at least one expert on orderless in this conversation. 😛

from consult.

@oantolin @s-kostyaev As far as I see, we don't have to do anything here? If you disagree, please reopen or ping me!

from consult.

I think if we will provide consult-isearch it should be aware of other isearch options like search-whitespace-regexp, isearch-lax-whitespace, isearch-regexp-lax-whitespace, etc. Also, it should use the same completion style as isearch. That's why I've created #40 to be able to overwrite default completion style for consult-line only.

from consult.

This one works for me:

  (defun my-consult-line-from-isearch ()
    (interactive)
    (minibuffer-with-setup-hook
	(lambda ()
	  (setq orderless-transient-matching-styles '(orderless-prefixes)))
      (consult-line-from-isearch)))

from consult.

Out of curiosity, @s-kostyaev: what styles do you normally use?

(orderless-matching-styles '(orderless-regexp orderless-initialism orderless-flex))

And for me flex breaks in-file search.

from consult.

I think consult-line should not overwrite isearch-string at all. It should have their own history (I mean consult-line). This command consult-line-from-isearch is great for add it to isearch-mode-map like this:

(leaf isearch
  :bind ((isearch-mode-map
	  ("M-i" . consult-line-from-isearch))))

or to use it with ace-isearch. But switching back and forward sounds neat. Name convention also from ace-isearch.

from consult.

@s-kostyaev I think overwriting would be okay since it would only happen for consult-isearch, thus being triggered explicitly by the user. I tried this. But so far the experiment did not go well. If you would like to play around with this a bit it could be cool, but we don't have to do this. It was just an idea. Thanks for the isearch-mode-map example. I agree, that makes sense. Maybe this should be added to the readme.

from consult.

@s-kostyaev Here is what I tried for now. As I said, does not work yet unfortunately.

;;;###autoload
(defun consult-isearch ()
  "Search by lines, using and updating the `isearch-string'."
  (interactive)
  (minibuffer-with-setup-hook
      (lambda ()
        (add-hook 'post-command-hook
                  (lambda ()
                    (setq
                     isearch-regexp nil
                     isearch-string (minibuffer-contents-no-properties))
         nil t)))
    (consult-line isearch-string)))

from consult.

Also, can consult-line search for regexp as isearch can?

from consult.

Also, can consult-line search for regexp as isearch can?

Depends on the completion style. If the completion style recognizes regexps, I think yes. But this is all a bit fuzzy. I am not deep into this. I am using orderless.

from consult.

I am using orderless.

Same as me. But looks like orderless has its own completion styles. And, AFAIK, that styles are global.

UPD. Yep. orderless has orderless-matching-styles. And it is a global custom variable.

from consult.

@s-kostyaev Here is what I tried for now. As I said, does not work yet unfortunately.

;;;###autoload
(defun consult-isearch ()
  "Search by lines, using and updating the `isearch-string'."
  (interactive)
  (minibuffer-with-setup-hook
      (lambda ()
        (add-hook 'post-command-hook
                  (lambda ()
                    (setq
                     isearch-regexp nil
                     isearch-string (minibuffer-contents-no-properties))
         nil t)))
    (consult-line isearch-string)))

This is a bad idea to use hook with lambda. You will add this lambda to that hook every time you call this command. Use named function instead.

from consult.

It is okay to do it like this here, but this is a matter of taste. And no, it is not added every time. The hook is added locally.

from consult.

Same as me. But looks like orderless has its own completion styles. And, AFAIK, that styles are global.

Probably you can overwrite them locally, like the hooks. But I would have to look into this.

from consult.

Yep, using setq-local on orderless-matching-styles should work.

from consult.

Yep, using setq-local on orderless-matching-styles should work.

Should I set it inside some hook?

from consult.

Well, you should set it when the minibuffer is the current buffer. So if you are already using minibuffer-with-setup-hook, do it in the function you pass to it.

from consult.

But, I for one don't like people messing with my matching styles, maybe others feel the same way.

I think a better solution is education: just explain clearly in the docstrings that consult-line uses whatever kind of completion you have and is not the same thing as isearch, regexp isearch or lax whitespace isearch.

from consult.

I prefer different completion styles for in file search and commands for example. I like Emacs because I can make it my own tool with my own rules. So why should I learn the tool instead of configure it for exact my purpose? Kind of philosophical question.

from consult.

Yep, using setq-local on orderless-matching-styles should work.

There is the better solution - orderless-transient-matching-styles

from consult.

@oantolin

But, I for one don't like people messing with my matching styles, maybe others feel the same way.
I think a better solution is education: just explain clearly in the docstrings that consult-line uses whatever kind of completion you have and is not the same thing as isearch, regexp isearch or lax whitespace isearch.

I agree with this sentiment, but such behavior could be made configurable and it could be restricted to only few commands for which this is an explicitly stated feature. But I am not sure if something like consult-isearch can be made or makes much sense. As I said it was just an idea based on the simpler command we already have consult-line-from-isearch.

from consult.

I think I can now achieve the behavior I need. Maybe consult-line-frome-isearch is enough.

from consult.

Out of curiosity, @s-kostyaev: what styles do you normally use?

from consult.

I see, that makes sense.

from consult.

I'm new to Consult, but I don't see the solution -- how to I connect consult to my isearch results when I press M-i if I'm not using orderless?

from consult.

@WorldsEndless What do you want to achieve exactly? You can access the last isearch input in the future history of consult-line. This is pretty much it, very limited. But there is also the proposal to make the isearch history accessible, see #171. But with #171, the isearch history would not connect to consult-line, but would only allow to restart an old isearch session.

from consult.

@WorldsEndless For reference, regarding ace-isearch I answered here - https://www.reddit.com/r/emacs/comments/l2sata/finding_replacements_for_switching_to_selectrum/gk80veh?utm_source=share&utm_medium=web2x&context=3

from consult.