helm-ff-checksum action should not be available for directories

[xiongtx]

Expected behavior

helm-ff-checksum action, found in the actions list of commands like helm-find-files, should not be available for directories.

Actual behavior from emacs-helm.sh if possible (See note at bottom)

helm-ff-checksum is available for directories. Calling it signals an error:

error in process sentinel: Read error: Is a directory, <dir-name>

but not before prompting the user for the checksum algorithm, which is unnecessary work.

Steps to reproduce (recipe)

1. Call helm-find-files
2. Move selection to directory
3. Call helm-ff-checksum (<f4>)

Backtraces if some (M-x toggle-debug-on-error)

Debugger entered--Lisp error: (file-error "Read error" "Is a directory" "/home/txx/.emacs.d/elpa/helm-20170511.649")
  signal(file-error ("Read error" "Is a directory" "/home/txx/.emacs.d/elpa/helm-20170511.649"))
  async-handle-result(#[257 "\302�!\210\303\304\300\301#\207" [sha256 "helm-20170511.649" kill-new message "Calculating %s checksum for `%s' done and copied to kill-ring"] 5 "\n\n(fn SUM)"] (async-signal (file-error "Read error" "Is a directory" "/home/txx/.emacs.d/elpa/helm-20170511.649")) #<buffer *emacs*>)
  async-when-done(#<process emacs> "finished\n")

Describe versions of helm, emacs, operating system etc.

• Helm: 20170511.649
• Emacs: 25.1.1
• OS: Fedora 24

[thierryvolpiatto]

Done, thanks.

[xiongtx]

Here's a thought: optional actions should come last, not be injected into the middle.

If we have actions like:

<f1> A
<f2> B
<f3> C

It makes more sense to append an action like:

<f1> A
<f2> B
<f3> C
<f4> D

than inject it like:

<f1> A
<f2> B
<f3> D
<f4> C

because injecting an action shifts the keys for every subsequent action; this requires the user to remember that C is <f4> for files, for example, but <f3> for dirs.

This maybe only applies to actions bound to keys, i.e. <f1>-<f12>.

[thierryvolpiatto]

Tianxiang Xiong <[email protected]> writes:

Here's a thought: optional actions should come last, not be injected into the middle. because injecting an action shifts the keys for every subsequent action; this requires the user to remember that C is <f4> for files, for example, but <f3> for dirs.

This is a good reason to move optional actions at bottom but there is also good reasons to move some actions on top... So, well I opted for moving actions on top or nearly because generally these actions are really what you need on this kind of candidate. Also this allow user to discover a new action have been added, at bottom they don't see it. But perhaps it is better moving all at bottom except the ones affecting first action e.g ~/.emacs.d/emacs-config/.emacs.el! C-x C-f RET and ~/.emacs.d/emacs-config/.emacs.el:754! C-x C-f RET don't know for now...

…

-- Thierry Gpg Key fingerprint = 6CEC 7081 AB33 E251 4AB8 5FC2 28D1 7F53 59F2 9997