bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Hi,

if I do this from emacs -Q:

- M-x fido-vertical-mode
- C-x C-f
- Type /ssh:
- Press TAB

I see the *Completions* buffer with this content:

Click on a completion to select it.
In this buffer, type RET to select the completion near point.

Possible completions are:
ssh:
sshx:
sshfs:

If I instead do not enable fido-vertical-mode, and I do:

- C-x C-f
- Type /ssh:
- Press TAB

I see this in the *Completions* buffer:

Click on a completion to select it.
In this buffer, type RET to select the completion near point.

Possible completions are:
ssh:work-serv:
ssh:server:

Which are the hosts coming from my ~/.ssh/config, and this is what I was 
expecting with fido-vertical-mode.

I noticed fido-vertical-mode uses icomplete--fido-mode-setup, which among other 
things sets (with setq-local) completion-styles to '(flex).

Now, if I use icomplete-vertical-mode with completion-styles set to
'(basic partial-completion flex), pressing TAB after /ssh: shows me the hosts I 
was looking for.


In GNU Emacs 29.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version 3.24.18, cairo 
version 1.16.0)
  of 2021-10-25 built on hathaway
Repository revision: 79f7e87da5037f22be07954bb8000ee88e18e515
Repository branch: master
Windowing system distributor 'The X.Org Foundation', version 11.0.12009000
System Description: Ubuntu 20.04 LTS

Configured using:
  'configure --with-harfbuzz --with-native-compilation'

Configured features:
ACL CAIRO DBUS FREETYPE GIF GLIB GMP GNUTLS GPM GSETTINGS HARFBUZZ JPEG
JSON LCMS2 LIBOTF LIBSELINUX LIBSYSTEMD LIBXML2 M17N_FLT MODULES
NATIVE_COMP NOTIFY INOTIFY PDUMPER PNG RSVG SECCOMP SOUND THREADS TIFF
TOOLKIT_SCROLL_BARS X11 XDBE XIM XPM GTK3 ZLIB

Important settings:
   value of $LC_MESSAGES: en_GB.UTF-8
   value of $LC_MONETARY: it_IT.UTF-8
   value of $LC_NUMERIC: it_IT.UTF-8
   value of $LC_TIME: it_IT.UTF-8
   value of $LANG: en_US.UTF-8
   value of $XMODIFIERS: @im=ibus
   locale-coding-system: utf-8-unix

Major mode: Lisp Interaction

Minor modes in effect:
   shell-dirtrack-mode: t
   icomplete-mode: t
   tooltip-mode: t
   global-eldoc-mode: t
   eldoc-mode: t
   show-paren-mode: t
   electric-indent-mode: t
   mouse-wheel-mode: t
   tool-bar-mode: t
   menu-bar-mode: t
   file-name-shadow-mode: t
   global-font-lock-mode: t
   font-lock-mode: t
   blink-cursor-mode: t
   auto-composition-mode: t
   auto-encryption-mode: t
   auto-compression-mode: t
   line-number-mode: t
   indent-tabs-mode: t
   transient-mark-mode: t

Load-path shadows:
None found.

Features:
(shadow sort mail-extr emacsbug message rmc puny dired dired-loaddefs
rfc822 mml mml-sec epa derived epg rfc6068 epg-config gnus-util rmail
rmail-loaddefs text-property-search mm-decode mm-bodies mm-encode
mail-parse rfc2231 mailabbrev gmm-utils mailheader sendmail rfc2047
rfc2045 ietf-drums mm-util mail-prsvr mail-utils tramp-cache tramp
tramp-loaddefs trampver tramp-integration files-x tramp-compat shell
pcomplete comint ansi-color ring parse-time iso8601 time-date ls-lisp
format-spec auth-source eieio eieio-core eieio-loaddefs password-cache
json map cus-start comp comp-cstr warnings rx cl-seq cl-macs cl-extra
seq gv subr-x byte-opt bytecomp byte-compile cconv cus-load icomplete
help-mode cl-loaddefs cl-lib iso-transl tooltip eldoc paren electric
uniquify ediff-hook vc-hooks lisp-float-type elisp-mode mwheel
term/x-win x-win term/common-win x-dnd tool-bar dnd fontset image
regexp-opt fringe tabulated-list replace newcomment text-mode lisp-mode
prog-mode register page tab-bar menu-bar rfn-eshadow isearch easymenu
timer select scroll-bar mouse jit-lock font-lock syntax font-core
term/tty-colors frame minibuffer cl-generic cham georgian utf-8-lang
misc-lang vietnamese tibetan thai tai-viet lao korean japanese eucjp-ms
cp51932 hebrew greek romanian slovak czech european ethiopic indian
cyrillic chinese composite emoji-zwj charscript charprop case-table
epa-hook jka-cmpr-hook help simple abbrev obarray cl-preloaded nadvice
button loaddefs faces cus-face macroexp files window text-properties
overlay sha1 md5 base64 format env code-pages mule custom widget
hashtable-print-readable backquote threads dbusbind inotify lcms2
dynamic-setting system-font-setting font-render-setting cairo
move-toolbar gtk x-toolkit x multi-tty make-network-process
native-compile emacs)

Memory information:
((conses 16 114626 9140)
  (symbols 48 10350 1)
  (strings 32 28393 2886)
  (string-bytes 1 959527)
  (vectors 16 19385)
  (vector-slots 8 325769 14133)
  (floats 8 45 290)
  (intervals 56 506 0)
  (buffers 992 12))

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

See bug#50387.  This is a recent "improvement".  Now you're supposed to 
type the first character of the hostname to see completions.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 25/10/21 16:54, Gregory Heytings wrote:
> See bug#50387.  This is a recent "improvement".  Now you're supposed to type the 
> first character of the hostname to see completions.

It doesn't seem to work, though. From emacs -Q:

- M-x fido-vertical-mode
- C-x C-f
- /ssh:s
- press TAB

The hostname isn't completed and I don't see the *Completions* buffer. While 
without enabling fido-vertical-mode, I get /ssh:server on pressing TAB after /ssh:s

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

[-- Attachment #1: Type: text/plain, Size: 669 bytes --]


>
> It doesn't seem to work, though. From emacs -Q:
>

Tramp doesn't behave as expected with only flex in completion-styles, 
indeed.  I see the same problem in Emacs 27.  Your bug report was similar 
to bug#50387, it is in fact related but subtly different.

>
> The hostname isn't completed and I don't see the *Completions* buffer. 
> While without enabling fido-vertical-mode, I get /ssh:server on pressing 
> TAB after /ssh:s
>

I'd suggest to add something a fallback method in 
icomplete--fido-mode-setup, for example completion-styles '(flex basic). 
And perhaps to apply the attached patch in your local Emacs, which 
provides I think the behavior you expect.

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Type: text/x-diff; name=Improve-handling-of-non-default-completion-styles-wi.patch, Size: 1089 bytes --]

From 5f4b821bbf5123422d43895bee4c847cfadf31b0 Mon Sep 17 00:00:00 2001
From: Gregory Heytings <gregory@heytings.org>
Date: Mon, 6 Sep 2021 20:52:21 +0000
Subject: [PATCH] Improve handling of non-default completion styles with Tramp
 methods

* lisp/net/tramp.el (tramp-completion-handle-file-name-all-completions):
Do not return methods when the filename already contains a complete method.
Fixes bug#50387.
---
 lisp/net/tramp.el | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/lisp/net/tramp.el b/lisp/net/tramp.el
index 4fd7a322d4..259fc1fcfc 100644
--- a/lisp/net/tramp.el
+++ b/lisp/net/tramp.el
@@ -2839,8 +2839,9 @@ tramp-completion-handle-file-name-all-completions
 			       (delq nil all-user-hosts)))))
 
 	    ;; Possible methods.
-	    (setq result
-		  (append result (tramp-get-completion-methods m)))))))
+	    (unless (string-empty-p method)
+	      (setq result
+		    (append result (tramp-get-completion-methods m))))))))
 
     ;; Unify list, add hop, remove nil elements.
     (dolist (elt result)
-- 
2.33.0

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 25/10/21 18:13, Gregory Heytings wrote:
> I'd suggest to add something a fallback method in icomplete--fido-mode-setup, 
> for example completion-styles '(flex basic). And perhaps to apply the attached 
> patch in your local Emacs, which provides I think the behavior you expect.

FWIW, it works with '(basic partial-completion flex), even without your patch 
applied.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>> I'd suggest to add something a fallback method in 
>> icomplete--fido-mode-setup, for example completion-styles '(flex 
>> basic). And perhaps to apply the attached patch in your local Emacs, 
>> which provides I think the behavior you expect.
>
> FWIW, it works with '(basic partial-completion flex), even without your 
> patch applied.
>

Of course: if you prioritize simpler completion mechanisms, it works 
without the fix for more complex completion mechanisms.

Can this bug be closed from your POV?

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 26/10/21 09:13, Gregory Heytings wrote:
> Of course: if you prioritize simpler completion mechanisms, it works without the 
> fix for more complex completion mechanisms.
> 
> Can this bug be closed from your POV?

Not unless there is a documented way to work around this problem with fido-mode, 
IMO. The used should be told to customize completion-styles properly, I guess, 
provided it customization does not get overridden by icomplete--fido-mode-setup.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 26/10/21 17:30, Manuel Uberti wrote:
> On 26/10/21 09:13, Gregory Heytings wrote:
>> Of course: if you prioritize simpler completion mechanisms, it works without 
>> the fix for more complex completion mechanisms.
>>
>> Can this bug be closed from your POV?
> 
> Not unless there is a documented way to work around this problem with fido-mode, 
> IMO. The used should be told to customize completion-styles properly, I guess, 
> provided it customization does not get overridden by icomplete--fido-mode-setup.
> 

"used" was "user" and "it customization" was "its customization". Sorry about that.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Manuel Uberti via "Bug reports for GNU Emacs, the Swiss army knife of
text editors" <bug-gnu-emacs@gnu.org> writes:

Hi Manuel,

>>> Can this bug be closed from your POV?
>> Not unless there is a documented way to work around this problem
>> with fido-mode, IMO. The used should be told to customize
>> completion-styles properly, I guess, provided it customization does
>> not get overridden by icomplete--fido-mode-setup.
>> 
>
> "used" was "user" and "it customization" was "its customization". Sorry about that.

The Tramp manual says

--8<---------------cut here---------------start------------->8---
User name and host name completion is activated only, if file name
completion has one of the styles ‘basic’, ‘emacs21’, or ‘emacs22’.
--8<---------------cut here---------------end--------------->8---

Furthermore, it mentions the problem of typing at least the first letter
of a host or user name in a footnote, see (info "(tramp) File name completion-Footnote-1")

--8<---------------cut here---------------start------------->8---
   (1) Some completion styles, like ‘substring’ or ‘flex’, require to
type at least one character after the trailing ‘:’.
--8<---------------cut here---------------end--------------->8---

Granted, this is very short. Proposals for better wording in the Tramp
manual are welcome.

Bonus point, if somebody who knows completion styles by heart could fix
this annoyance. Then we don't need to document it.

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> Bonus point, if somebody who knows completion styles by heart could fix 
> this annoyance. Then we don't need to document it.
>

??? I sent you a (one-line!) patch for the second annoyance, and you 
rejected it.  I'd be willing to look at the first annoyance, but not if 
it's to see the patch again rejected.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Gregory,

>> Bonus point, if somebody who knows completion styles by heart could
>> fix this annoyance. Then we don't need to document it.
>>
>
> ??? I sent you a (one-line!) patch for the second annoyance, and you
> rejected it.  I'd be willing to look at the first annoyance, but not
> if it's to see the patch again rejected.

Your patch violates the file-name-all-completions API. Testing with
"emacs -Q -l tramp":

Currently, we have the following with vanilla Emacs (on my laptop):

--8<---------------cut here---------------start------------->8---
(file-name-all-completions "" "/")
=> ("sudoedit:" "sshfs:" "smb:" "rcp:" "remcp:" "scp:" "scpx:" "rsync:" "rsh:" "remsh:" "ssh:" "sshx:" "telnet:" "nc:" "su:" "sg:" "sudo:" "doas:" "ksu:" "krlogin:" "plink:" "plinkx:" "pscp:" "psftp:" "fcp:" "rclone:" "afp:" "dav:" "davs:" "gdrive:" "mtp:" "nextcloud:" "sftp:" "ftp:" "adb:" "sbin/" "proc/" "bin/" "snap/" "opt/" "var/" "dev/" "lib64/" "./" "nonexistent/" "lost+found/" "media/" "root/" "srv/" "etc/" "mnt/" "../" "tmp/" "net/" "sys/" "lib/" "boot/" "home/" "run/" "usr/")
--8<---------------cut here---------------end--------------->8---

With your one-line patch, we have instead

--8<---------------cut here---------------start------------->8---
(file-name-all-completions "" "/")
=> ("sbin/" "proc/" "bin/" "snap/" "opt/" "var/" "dev/" "lib64/" "./" "nonexistent/" "lost+found/" "media/" "root/" "srv/" "etc/" "mnt/" "../" "tmp/" "net/" "sys/" "lib/" "boot/" "home/" "run/" "usr/")
--8<---------------cut here---------------end--------------->8---

That means, all Tramp methods are suppressed. This is wrong, a valid
file name can start with a Tramp method. The Elisp manual calls such
cnstructs "magic" file names, but they are still valid file name
prefixes in the sense of file-name-all-completions. And so they are
valid results.

Our discrepancy is, that I believe we shall follow the API, and fix
problems where they are located. You want a workaround in Tramp, which
would IMO do more damage than it is good for. My experience tells me
that such hacks will create new problems we have to deal with in the
future.

Both of us agree that there is a problem. I don't intend to reject
patches from you, I just must be convinced that they solve the problem
adequately.

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> Your patch violates the file-name-all-completions API. Testing with 
> "emacs -Q -l tramp":
>

We've discussed this earlier.  I concluded that the problem is in the 
file-name-all-completions, where it can be fixed with a single line 
change.  You believe that the problem should be fixed at the level of the 
completion mechanisms, which won't happen.

>
> Currently, we have the following with vanilla Emacs (on my laptop):
>
> (file-name-all-completions "" "/") => ("sudoedit:" ...)
>
> With your one-line patch, we have instead
>
> (file-name-all-completions "" "/") => ("sbin/" ...)
>
> That means, all Tramp methods are suppressed.
>

That's a very partial way of presenting what my patch does.  The truth is 
that Tramp methods are not at all suppressed.  They are not returned when 
the completions of the root directory are required, they are returned as 
soon as at least one character is typed:

(file-name-all-completions "s" "/") => ("sudoedit:" "sshfs:" ... "sys/" ...)

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 26/10/21 20:12, Michael Albinus wrote:
> Furthermore, it mentions the problem of typing at least the first letter
> of a host or user name in a footnote, see (info "(tramp) File name completion-Footnote-1")
> 
> --8<---------------cut here---------------start------------->8---
>     (1) Some completion styles, like ‘substring’ or ‘flex’, require to
> type at least one character after the trailing ‘:’.
> --8<---------------cut here---------------end--------------->8---

As I reported before, at the moment from 'emacs -Q' and with fido-vertical-mode 
enabled, typing one character after the trailing ':' does not offer the 
completions I can see without fido-vertical-mode.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

[-- Attachment #1: Type: text/plain, Size: 774 bytes --]


>
> As I reported before, at the moment from 'emacs -Q' and with 
> fido-vertical-mode enabled, typing one character after the trailing ':' 
> does not offer the completions I can see without fido-vertical-mode.
>

In your case both conditions apply:

1. User name and host name completion is activated only, if file name 
completion has one of the styles ‘basic’, ‘emacs21’, or ‘emacs22’.

2. Some completion styles, like ‘substring’ or ‘flex’, require to type at 
least one character after the trailing ‘:’.

The second condition could perhaps be reworded as "Some completion styles, 
like 'substring' or 'flex', require to type at least one character after 
the trailing ':' *when they come first in the completion-styles list*."

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 27/10/21 09:51, Gregory Heytings wrote:
> The second condition could perhaps be reworded as "Some completion styles, like 
> 'substring' or 'flex', require to type at least one character after the trailing 
> ':' *when they come first in the completion-styles list*."

Just wondering: wouldn't be enough to change icomplete--fido-mode-setup to use:

(setq-local completion-styles '(flex basic))

Instead of:

(setq-local completion-styles '(flex))

As it is doing currently?

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>> The second condition could perhaps be reworded as "Some completion 
>> styles, like 'substring' or 'flex', require to type at least one 
>> character after the trailing ':' *when they come first in the 
>> completion-styles list*."
>
> Just wondering: wouldn't be enough to change icomplete--fido-mode-setup 
> to use:
>
> (setq-local completion-styles '(flex basic))
>
> Instead of:
>
> (setq-local completion-styles '(flex))
>
> As it is doing currently?
>

It depends what you mean by "being enough".  That setting would be better, 
but you would still have to type the first character of the hostname. 
With that setting and my patch, everything would work as you would expect, 
namely, you wouldn't have to type the first character of the hostname 
anymore.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 27/10/21 10:05, Gregory Heytings wrote:
> It depends what you mean by "being enough".  That setting would be better, but 
> you would still have to type the first character of the hostname. With that 
> setting and my patch, everything would work as you would expect, namely, you 
> wouldn't have to type the first character of the hostname anymore.

I see. Well then yes, my opinion is that both the completion-styles setting and 
your patch should be added.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Gregory,

>>> The second condition could perhaps be reworded as "Some completion
>>> styles, like 'substring' or 'flex', require to type at least one
>>> character after the trailing ':' *when they come first in the
>>> completion-styles list*."
>>
>> Just wondering: wouldn't be enough to change
>> icomplete--fido-mode-setup to use:
>>
>> (setq-local completion-styles '(flex basic))
>>
>> Instead of:
>>
>> (setq-local completion-styles '(flex))
>>
>> As it is doing currently?
>
> It depends what you mean by "being enough".  That setting would be
> better, but you would still have to type the first character of the
> hostname. With that setting and my patch, everything would work as you
> would expect, namely, you wouldn't have to type the first character of
> the hostname anymore.

The better setting is '(basic flex). With that setting, there is no
problem at all. Try

emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))"

This works out-of-the box, w/o any further patch (as you have said above).

If I do instead

emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))" -f fido-vertical-mode

it doesn't work. Debugging of fido-vertical-mode is needed, I suppose.

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 27/10/21 11:05, Michael Albinus wrote:
> If I do instead
> 
> emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))" -f fido-vertical-mode
> 
> it doesn't work. Debugging of fido-vertical-mode is needed, I suppose.

Hi Michael,

as I wrote before, I think it's because fido-vertical-mode uses 
icomplete--fido-mode-setup, which locally sets completion-styles:

(defun icomplete--fido-mode-setup ()
   "Setup `fido-mode''s minibuffer."
   (when (and icomplete-mode (icomplete-simple-completing-p))
     (use-local-map (make-composed-keymap icomplete-fido-mode-map
                                          (current-local-map)))
     (setq-local icomplete-tidy-shadowed-file-names t
                 icomplete-show-matches-on-no-input t
                 icomplete-hide-common-prefix nil
                 icomplete-scroll (not (null icomplete-vertical-mode))
                 completion-styles '(flex)
                 completion-flex-nospace nil
                 completion-category-defaults nil
                 completion-ignore-case t
                 read-buffer-completion-ignore-case t
                 read-file-name-completion-ignore-case t)))

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> The better setting is '(basic flex). With that setting, there is no 
> problem at all. Try
>
> emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))"
>

That's better from your point of view.  Some users do not want to use the 
"basic" completion backend before the "flex" completion backend, they want 
to use "flex" and, only when "flex" fails, "basic".

>
> If I do instead
>
> emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))" -f 
> fido-vertical-mode
>
> it doesn't work. Debugging of fido-vertical-mode is needed, I suppose.
>

This has already been done, as Manuel suggested the solution would be to 
add a fallback in icomplete--fido-mode-setup, which amounts to doing (set 
completion-styles '(flex basic)).

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Dmitry Gutov]

On 27.10.2021 10:55, Manuel Uberti via Bug reports for GNU Emacs, the 
Swiss army knife of text editors wrote:
> (setq-local completion-styles '(flex basic))

That looks a little silly: flex should match anywhere where basic matches.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>> (setq-local completion-styles '(flex basic))
>
> That looks a little silly: flex should match anywhere where basic 
> matches.
>

Perhaps it should, but it doesn't, especially when the completion provider 
tries to be smart.  See:

emacs -Q -l tramp --eval "(setq completion-styles '(flex))"
C-x C-f /ssh: TAB

emacs -Q -l tramp --eval "(setq completion-styles '(basic))"
C-x C-f /ssh: TAB

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Dmitry Gutov]

On 27.10.2021 13:23, Gregory Heytings wrote:
> 
> Perhaps it should, but it doesn't, especially when the completion 
> provider tries to be smart.

I'm not sure "smart" is the word I'd use.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Grepor,

>> If I do instead
>>
>> emacs -Q -l tramp --eval "(setq completion-styles '(basic flex))" -f
>> fido-vertical-mode
>>
>> it doesn't work. Debugging of fido-vertical-mode is needed, I suppose.
>>
>
> This has already been done, as Manuel suggested the solution would be
> to add a fallback in icomplete--fido-mode-setup, which amounts to
> doing (set completion-styles '(flex basic)).

Yep. However, even if I apply this patch to icomplete.el and also your
Tramp patch it doesn't complete method names like "/ss TAB". So this
cannot be the proper solution I believe.

And, beside of fido-vertical-mode, your Tramp patch hides the method
names at all. If I try

emacs -Q -l tramp --eval "(setq completion-styles '(flex basic))"

and apply then 'C-x C-f /ss TAB', it offers only the candidate '/sys/',
and nothing else. W/o your patch, there are the candidates

--8<---------------cut here---------------start------------->8---
Possible completions are:
sshfs: 	ssh:
sshx: 	sys/
--8<---------------cut here---------------end--------------->8---

Anyway, all our tests are mood. I would really appreciate if a
completion style author could chime in, and could investigate. I don't
say that there isn't an error in Tramp, but before I accept patches I
must understand what's up. And the proposed patch doesn't fix all cases
sufficiently (plus it violates the API, but this we have discussed
already in several rounds).

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Dmitry Gutov]

On 27.10.2021 14:10, Michael Albinus wrote:
> Anyway, all our tests are mood. I would really appreciate if a
> completion style author could chime in, and could investigate. I don't
> say that there isn't an error in Tramp, but before I accept patches I
> must understand what's up. And the proposed patch doesn't fix all cases
> sufficiently (plus it violates the API, but this we have discussed
> already in several rounds).

flex aside, does this completion work with the 'substring' completion 
style, for example?

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 27/10/21 13:19, Dmitry Gutov wrote:
> On 27.10.2021 14:10, Michael Albinus wrote:
>> Anyway, all our tests are mood. I would really appreciate if a
>> completion style author could chime in, and could investigate. I don't
>> say that there isn't an error in Tramp, but before I accept patches I
>> must understand what's up. And the proposed patch doesn't fix all cases
>> sufficiently (plus it violates the API, but this we have discussed
>> already in several rounds).
> 
> flex aside, does this completion work with the 'substring' completion style, for 
> example?

 From emacs -Q:

- (setq completion-styles '(substring))
- C-x C-f
- /ssh
- TAB

I get the *Completions* buffer with this content:

Click on a completion to select it.
In this buffer, type RET to select the completion near point.

Possible completions are:
ssh:
sshfs:
sshx:

While with the following from emacs -Q:

- (setq completion-styles '(substring))
- C-x C-f
- /ssh:
- TAB

I don't get any *Completions* buffer at all.

-- 
Manuel Uberti
www.manueluberti.eu

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> And, beside of fido-vertical-mode, your Tramp patch hides the method 
> names at all.
>

No, it does not.

>
> If I try
>
> emacs -Q -l tramp --eval "(setq completion-styles '(flex basic))"
>
> and apply then 'C-x C-f /ss TAB', it offers only the candidate '/sys/', 
> and nothing else. W/o your patch, there are the candidates
>

Yes, this is because "flex" is tried first, and returns "sys" because it 
contains two "s".

With that setting (flex basic), everything works perfectly with C-x C-f 
/ssh TAB, after which you see "ssh: sshfs: sshx:", and with C-x C-f /ssh: 
TAB, after which you see the hostnames.

With (substring basic), C-x C-f /ss TAB is completed and becomes /ssh.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Gregory,

>> That looks a little silly: flex should match anywhere where basic
>> matches.
>
> Perhaps it should, but it doesn't, especially when the completion
> provider tries to be smart.  See:
>
> emacs -Q -l tramp --eval "(setq completion-styles '(flex))"
> C-x C-f /ssh: TAB
>
> emacs -Q -l tramp --eval "(setq completion-styles '(basic))"
> C-x C-f /ssh: TAB

Agreed. Should be fixed first.

And this is what the Tramp manual says. I've grepped the archives, and
it seems I've added this sentence after a discussion in the thread
starting at <https://lists.gnu.org/archive/html/help-gnu-emacs/2019-10/msg00129.html>.
There's an interesting contribution from Stefan Monnier, but all what he
has said is out of my scope. I don't know what to do with completion
styles.

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> flex aside, does this completion work with the 'substring' completion 
> style, for example?
>

It depends what you mean by "work" ;-)

1. Without my patch:

1.1. With (substring) alone:

/ TAB lists all Tramp methods and all directories in the root directory

/ ss TAB is completed and offers the ssh: sshfs: sshx: methods

/ ssh: TAB displays "Sole completion", and it's not possible to complete 
hostnames (with or without a first char)

1.2. With (substring basic):

/ TAB lists all Tramp methods and all directories in the root directory

/ ss TAB is completed and offers the ssh: sshfs: sshx: methods

/ ssh: TAB displays "Sole completion", but after typing one more char you 
get the list of hostnames which start with that char

2. With my patch:

2.1. With (substring) alone, nothing works, Tramp methods and hostnames 
are not recognized, which is at least coherent with what the manual says: 
"User name and host name completion is activated only, if file name 
completion has one of the styles 'basic', 'emacs21', or 'emacs22'".

2.2. With (substring basic):

/ TAB lists all directories in the root directory

/ ss TAB is completed and offers the ssh: sshfs: sshx: methods

/ ssh: TAB displays the list of hostnames

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Gregory,

>> And, beside of fido-vertical-mode, your Tramp patch hides the method
>> names at all.
>
> No, it does not.

I'm pretty sure we won't find a solution if we continue to repeat our
arguments.

The message from Stefan (see the reference in my other mail) seems to be
promising. I would like to go the way he proposes, but I'm a dummy in
completion styles.

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

Hi Gregory,

> 2. With my patch:
>
> 2.1. With (substring) alone, nothing works, Tramp methods and
> hostnames are not recognized, which is at least coherent with what the
> manual says: "User name and host name completion is activated only, if
> file name completion has one of the styles 'basic', 'emacs21', or
> 'emacs22'".

Yes. But this statement in the Tramp manual isn't a feature. It rather
explains an error, and how to work around. So no, (substring) alone
doesn't "work".

Best regards, Michael.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> Agreed. Should be fixed first.
>

I don't know what kind of fix you expect, but FTR with my patch everyting 
works as expected with the following completion-styles:

(basic partial-completion emacs22) ;; default
(initials basic)
(flex basic)
(substring basic)

Without, only the two first work as expected.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Gregory Heytings]

>
> The message from Stefan (see the reference in my other mail) seems to be 
> promising.
>

I'll have a look.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Michael Albinus]

Gregory Heytings <gregory@heytings.org> writes:

>> The message from Stefan (see the reference in my other mail) seems
>> to be promising.
>
> I'll have a look.

Thank you.

bug#51386: 29.0.50; fido-vertical-mode: ssh host completion

[Manuel Uberti via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Just to add another pair of eyes on this, I am CC'ing João Távora here because I 
know he put some work on fido-mode and maybe could share his opinion on this.

-- 
Manuel Uberti
www.manueluberti.eu