LaTeX-insert-item problem with end of line in evil mode

[brabalan]

Description

When using LaTeX-insert-item (,i in latex,

spacemacs/layers/+lang/latex/packages.el

Line 93 in f49792e

"i" 'LaTeX-insert-item ;; C-c C-j

), the character under the cursor is put in the following item. As it's not possible to go right of this character, on has to make sure there is a space at the end of the line before using ,i

Reproduction guide

• Start Emacs
• open a latex file
• insert an itemize environment ,e<RET>
• type afoo, then hit escape
• type ,i

Observed behaviour:
This is what we get:

\begin{itemize}
\item fo
\item o
\end{itemize}

Expected behaviour:
This is what we should get

\begin{itemize}
\item foo
\item 
\end{itemize}

System Info

• OS: darwin
• Emacs: 24.5.1
• Spacemacs: 0.105.11
• Spacemacs branch: master (rev. 93e55bb)
• Distribution: spacemacs
• Editing style: vim
• Completion: helm
• Layers:

(auto-completion elfeed emacs-lisp finance git gnus latex ocaml org
                 (spell-checking :variables spell-checking-enable-auto-dictionary t))

Backtrace

not relevant

[justbur]

I agree. I'll take this one.

[justbur]

try this

      (defun spacemacs/latex-insert-item ()
        (interactive "*")
        (if (evil-normal-state-p)
            (progn (evil-open-below 1)
                   (beginning-of-line)
                   (call-interactively 'LaTeX-insert-item))
          (call-interactively 'LaTeX-insert-item)))
      (spacemacs/set-leader-keys-for-major-mode 'latex-mode
        "i" 'spacemacs/latex-insert-item)

[brabalan]

It works great, thanks.

[dl3br]

if you use autocompletion layer there's a snippet bound on "-" (hyphen) for that.
Thus:
"- M-/" or "- TAB" or "- RET" depending on your config

[brabalan]

@dlebrecht thank you for the suggestion. It works with M-/ here, which is not very convenient.

[dl3br]

if that interests you, adding this to ur .spacemacs layer configuration, it
should work using tab key:
instead of simply auto-completion, you set variables

 (auto-completion :variables
                  auto-completion-enable-snippets-in-popup t
                  auto-completion-tab-key-behavior 'complete
                  )

On Thu, Feb 25, 2016 at 4:21 PM, Alan Schmitt notifications@github.com
wrote:

@dlebrecht https://github.com/dlebrecht thank you for the suggestion.
It works with M-/ here, which is not very convenient.

—
Reply to this email directly or view it on GitHub
#5241 (comment)
.

[brabalan]

@dlebrecht Thanks for the hint, but I think I found another bug, which I'll report separately (done as #5259): one-letter snippets do not work, as for some reason the completion mechanism is not triggered for them. As soon as I enter two letters, I see both completions from company (usual words), and snippets.

[ghost]

This is actually not LaTeX specific. This is rather a more general bug and issue with emacs in general. Since emacs is built around the idea of non-modal editing, it has no concept of modal editing. Thus, any command that acts to modify the state of the environment will do so with a "one character off" error. Making a new bullet point in org mode breaks the same way as inserting a new item does in LaTeX, for example. Anything sensitive to the position of the cursor will break this way.

The easiest fix that I can think of is to have commands like $ go one character past the end of a line so that all of those things work as expected. Otherwise trying to fix each individual thing will be nearly impossible if you want them to implicitly cut the remainder of the line down to the next line.

[TheBB]

You can set evil-move-beyond-eol to t in your config if you want to be able to move past the end of the line.

[brabalan]

@TheBB I used so have that setting before switching to spacemacs. It may be nice to put it by default.

[TheBB]

We did have it as default for a while and it was quite contentious.

[brabalan]

Why? If there is a link to a discussion, I would really like to know what the problems could be with that setting.

[TheBB]

#2525

[brabalan]

Very interesting read, thanks. So my understanding of the conclusion there is that we need to fix every function that relies on the character being after another character, and the fix by @justbur is the way to go.

[justbur]

My interpretation of this bug report was that you were expecting this function to open up a new line below the current one, when what it actually does is insert a new line into the current line, splitting the line if the point is not at the end of of the line.

My "fix" is not trying to reposition the point. It was actually rewriting the function to do something different in normal mode (open a new line below).

I think it would be awful to try and manually "fix" functions by repositioning the point depending on context.

[dschrempf]

Hi,
I cannot see why this has been closed, I repeatedly run into this problem that functions which assume to be at the eol do not work as expected.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Please let us know if this issue is still valid!