Fixes: bug#19032, bug#43120
Previous fixes to bug#19032 introduced bugs in Fido mode. This fix
relies on a new command bound to RET.
* etc/NEWS (Miscellaneous): Mention icomplete-show-matches-on-no-input.
* lisp/icomplete.el (icomplete-show-matches-on-no-input): Add comment.
(icomplete-minibuffer-map): Rebind minibuffer-complete-and-exit to
icomplete-ret.
(icomplete-ret): New command.
This reverts:
- commit 585fe00557, which is titled
"Fix up previous icomplete-show-matches-on-no-input change"
- commit 1b8d369c38, which is titled
"Change icomplete-show-matches-on-no-input behavior".
A simpler, less intrusive fix follows shortly after this commit.
Fixes: bug#43222
Icomplete mode re-sorts candidates, bubbling the default to top if
it's found somewhere down the list. This is done according to two
criteria: exact match and prefix match. Before this fix, it didn't
take into account the possibility that the exact match for the default
would already be on top, and would incorrectly bubble a prefixing
completion down the list to the top. This commit fixes that.
* lisp/icomplete.el (icomplete--sorted-completions):
Rework. Recomment.
* lisp/icomplete.el (icomplete-show-matches-on-no-input): Doc fix.
(icomplete-completions): Set completion-content-when-empty.
* lisp/minibuffer.el (completion-content-when-empty): New variable.
(completion--complete-and-exit): Use it (bug#19032).
Based on a patch by Matthew Leach <matthew@mattleach.net>.
Reported by: Andrew Schwartzmeyer <andrew@schwartzmeyer.com>
* lisp/icomplete.el (icomplete-fido-backward-updir): Don't save
dir fragments to kill ring.
By default, the flex completion style _does_ match spaces.
Fixes: bug#40625
* lisp/icomplete.el (icomplete--fido-mode-setup): Force
completion-flex-nospace to nil.
* lisp/minibuffer.el (completion-flex-nospace): Default to nil.
* lisp/minibuffer.el (minibuffer--require-match): New variable.
(completing-read-default): Bind it to the REQUIRE-MATCH value.
(minibuffer-force-complete-and-exit): Consult it to allow input
with no matches when a match is not required (bug#38992).
* lisp/icomplete.el (icomplete-exhibit): Use it to render the
correct parens around matches.
Besides the adverse effect of delaying completions, it tripped up the
useful logic of icomplete-force-complete-and-exit in the case where a
default was available, but no completions calculated yet.
* lisp/icomplete.el (icomplete-fido-ret): Don't force calculation
of completions.
(cherry picked from commit 13778aa5be)
This moves the logic from the series of commits starting in the commit named:
Improve sorting of flex completion style with non-nil minibuffer-default
to lisp/icomplete.el, so far the only confirmed beneficiary of that
functionality.
* lisp/icomplete.el (icomplete--sorted-completions): Consider
minibuffer-default here.
* lisp/minibuffer.el (completion--flex-adjust-metadata): Simplify.
(cherry picked from commit 0d2a711dc9)
Notably C-x d (M-x dired) and C-x v d (M-x vc-dir) behaved quite
differently, having regular files as the default instead of ido's
usual "./".
* lisp/icomplete.el (icomplete--sorted-completions): New helper.
(icomplete-completions): Use it.
(cherry picked from commit ee914051fb)
When using this option and editing input, some transient situations
may arise that lead to file-name shadowing, but that shouldn't
necessarily lead to auto-delete behaviour, which will be suprising.
In '/foo/x/bar', if the user deletes the 'x', shadowing occurs, but
probably shouldn't. So, somewhat like ido-mode, only auto-tidy
shadowed file names if the user is inserting text at end of input.
* lisp/icomplete.el (icomplete-exhibit): Check this-command.
(icomplete-tidy-shadowed-file-names): Tweak docstring.
Otherwise we would lose things like C-r (isearch-backward) when doing
simple stuff like non-icomplete pp-eval-expression.
* lisp/icomplete.el (icomplete--fido-mode-setup): Restrict setup
to when icomplete-mode really kicks in.
* lisp/icomplete.el (icomplete-fido-kill)
(icomplete-fido-delete-char, icomplete-fido-ret)
(icomplete-fido-backward-updir): Rename from icomplete-magic-ido-*
versions.
(icomplete-fido-mode-map): Use new command names.
* lisp/icomplete.el (icomplete-completions): Don't use ellipsis when
truncating determ in fido-mode. Highlight the center part of
determ with icomplete-first-match.
* lisp/icomplete.el (rfn-eshadow): Require it.
(icomplete-tidy-shadowed-file-names): New variable.
(icomplete-exhibit): Use icomplete-tidy-shadowed-file-names.
If there is no minibuffer input, but the user has already cycled some
pre-calculated completions, we should be calling
minibuffer-force-complete-and-exit instead of
minibuffer-complete-and-exit. The former is guaranteed to be fast in
this situation and yields the desired "selected" completion, while the
latter will just give us the default, ignoring all the cycling of
icomplete-{forward|backward}-completions.
* lisp/icomplete.el (icomplete-force-complete-and-exit): Add
comments and fix for empty input but some completions calculated.
Sometimes, when finding files with icomplete-mode, backward-deleting
the previous word or sexp (to move up a directory) doesn't actually
refresh the file list of the new directory. Forcing redisplay in
icomplete-exhibit misteriously fixes the problem.
* lisp/icomplete.el (icomplete-exhibit): Add call to redisplay.
This tries (not particularly hard) to emulate Ido's magic C-k, C-d,
RET and DEL. For now, the new commands have to be explicitly added to
an active map by the user, using something like this:
(let ((imap icomplete-minibuffer-map))
(define-key imap (kbd "C-k") 'icomplete-magic-ido-kill)
(define-key imap (kbd "C-d") 'icomplete-magic-ido-delete-char)
(define-key imap (kbd "RET") 'icomplete-magic-ido-ret)
(define-key imap (kbd "DEL") 'icomplete-magic-ido-backward-updir))
* lisp/icomplete.el (icomplete-magic-ido-kill)
(icomplete-magic-ido-delete-char, icomplete-magic-ido-ret)
(icomplete-magic-ido-backward-updir): New commands.
Often, when using icomplete with icomplete-show-matches-on-no-input
and commands like C-h f, icomplete-exhibit will take a long time (like
several seconds) to show all the completions.
However, if there is a non-nil default value, like when the C-h f is
performed exactly on a function name, the minibuffer prompt will
reflect that immediately. If the user immediately presses RE
(icomplete-force-complete-and-exit) we take that to mean "use the
default" and avoid the long wait.
As an extra performance tweak, this commit also removes an unneeded
call to icomplete-exhibit in icomplete-minibuffer-setup.
* lisp/icomplete.el (icomplete-force-complete-and-exit): Short-circuit
possibility of non-nil minibuffer-default.
(icomplete-minibuffer-setup): No need to explicitly icomplete-exhibit
here since icomplete-post-command-hook will call it unconditionally.
The convention is that a file with Author: but not Maintainer:
means the author is a maintainer, which makes it confusing
when a file lists the same person as author and maintainer.
Avoid the confusion by removing the duplicate Maintainer: line.
Rather than let minibuffer-force-complete set up cycling and then undoing it,
better tell it directly not to setup cycling. Also be a bit more careful
to remove the transient map.
Additionally to bug#34077 and bug#34116, this also relates to bug#25644.
* lisp/minibuffer.el (completion--flush-all-sorted-completions):
Also take down the transient cycling map if applicable.
(minibuffer-force-complete): New arg dont-cycle.
Set completion-cycling to the actual function that takes down the
transient map rather than just t.
(minibuffer-force-complete-and-exit):
* lisp/icomplete.el (icomplete-force-complete): Use new dont-cycle arg.
Cycling after forcing a completion with C-M-i in icomplete can be
confusing, as it leaves rotated prospects in the minibuffer. In C-x
C-f, for example it is very difficult to understand if the prospects
refer to subdirectories of the directory being completed to, which
happens naturally when the completion is unique; or if they are a
cycled version of prospects that match the new completion pattern, in
case the completion happens to still match other items.
To resolve this confusion, never cycle with C-M-i in icomplete:
non-ambiguous cycling can be achieved with C-. and C-,
The former behaviour can still be restored with:
(define-key icomplete-minibuffer-map (kbd "C-M-i") 'minibuffer-force-complete)
* lisp/icomplete.el (icomplete-force-complete): New command.
(icomplete-minibuffer-map): Bind C-M-i to icomplete-force-complete.
