Using racket-mode to edit source in org-mode document causing 100% cpu

[bhrgunatha]

I'm back again with a different 100% scenario - ref previous issue #318

• Emacs 26.1
• racket-mode - melpa 20181206.329
• org-mode 9.2 (see below though)

racket-bug-report output:

``` ((emacs-version "26.1") (emacs-uptime "7 minutes, 44 seconds") (system-type gnu/linux) (major-mode racket-mode) (racket--el-source-dir "/home/bhrgunatha/.emacs.d/.cask/26.1/elpa/racket-mode-20181206.329/") (racket--rkt-source-dir "/home/bhrgunatha/.emacs.d/.cask/26.1/elpa/racket-mode-20181206.329/racket/") (racket-program "racket") (racket-command-port 55555) (racket-command-timeout 10) (racket-memory-limit 2048) (racket-error-context medium) (racket-retry-as-skeleton t) (racket-error-context medium) (racket-history-filter-regexp "\\`\\s *\\S ?\\S ?\\s *\\'") (racket-images-inline t) (racket-images-keep-last 100) (racket-use-repl-submit-predicate nil) (racket-images-system-viewer "display") (racket-pretty-print t) (racket-indent-curly-as-sequence t) (racket-indent-sequence-depth 0) (racket-pretty-lambda nil) (racket-smart-open-bracket-enable nil) (racket-module-forms "\\s(\\(?:module[*+]?\\|library\\)") (racket-logger-config ((cm-accomplice . warning) (GC . info) (module-prefetch . warning) (optimizer . info) (racket/contract . error) (sequence-specialization . info) (* . fatal)))) (enabled-minor-modes (auto-composition-mode) (auto-compression-mode) (auto-encryption-mode) (auto-fill-mode) (auto-save-mode) (company-mode) (delete-selection-mode) (display-line-numbers-mode) (electric-indent-mode) (file-name-shadow-mode) (font-lock-mode) (global-company-mode) (global-display-line-numbers-mode) (global-eldoc-mode) (global-font-lock-mode) (global-hl-line-mode) (global-visual-line-mode) (hs-minor-mode) (ivy-mode) (line-number-mode) (menu-bar-mode) (mouse-wheel-mode) (org-src-mode) (paredit-mode) (pcre-mode) (rainbow-delimiters-mode) (show-paren-mode) (tooltip-mode) (transient-mark-mode) (visual-line-mode) (yas-global-mode) (yas-minor-mode)) (disabled-minor-modes (abbrev-mode) (auto-fill-function) (auto-save-visited-mode) (blink-cursor-mode) (buffer-read-only) (cl-old-struct-compat-mode) (column-number-mode) (company-search-mode) (compilation-in-progress) (compilation-minor-mode) (compilation-shell-minor-mode) (completion-in-region-mode) (counsel-mode) (defining-kbd-macro) (dired-hide-details-mode) (display-time-mode) (doc-view-minor-mode) (eldoc-mode) (electric-layout-mode) (electric-pair-mode) (electric-quote-mode) (global-prettify-symbols-mode) (gnus-dead-summary-mode) (gnus-undo-mode) (hl-line-mode) (horizontal-scroll-bar-mode) (ido-everywhere) (image-minor-mode) (isearch-mode) (jit-lock-debug-mode) (mail-abbrevs-mode) (mml-mode) (next-error-follow-minor-mode) (org-bullets-mode) (org-cdlatex-mode) (outline-minor-mode) (overwrite-mode) (package-build-minor-mode) (paragraph-indent-minor-mode) (prettify-symbols-mode) (racket-check-syntax-mode) (rxt--read-pcre-mode) (rxt-global-mode) (rxt-mode) (server-mode) (sh-electric-here-document-mode) (show-smartparens-global-mode) (show-smartparens-mode) (size-indication-mode) (smartparens-global-mode) (smartparens-global-strict-mode) (smartparens-mode) (smartparens-strict-mode) (temp-buffer-resize-mode) (tool-bar-mode) (unify-8859-on-decoding-mode) (unify-8859-on-encoding-mode) (url-handler-mode) (use-hard-newlines) (view-mode) (visible-mode) (window-divider-mode) (xref-etags-mode)) ```

I use the org-plus-contrib package from their own repository http://orgmode.org/elpa which bundles org mode plus contributed packages.

To reproduce this issue:

• create a new file test.org
• create a racket source code block C-s C-, s TAB or just

#+begin_src racket

#+end_src

• Use racket-mode to edit the code block in a separate window C-c '
• In the new window type in (de| ... and a message appears in the mini-buffer immediately saying "Still trying to connect to the racket-command process on port 55555 but emacs becomes unresponsive at 100% CPU in a similar way to melpa racket-mode causes emacs to become unresponsive with 100% CPU #318

A couple of observations:
Any typing at the keyboard is inserted into the buffer after C-g is pressed and it seems C-g works consistently to end the problem.
Making a further source code block works fine and doesn't hang and the racket REPL buffer is also still present.

I think this is a racket-mode issue rather than org-mode because editing a python code block doesn't prevent company mode completion or cause a CPU spike.

[greghendershott]

Thank you for the report.

This might be some kind of variation of #344 which also involves company-mode.

---

Although I think company-mode is wonderful for auto-completion, I prefer pressing TAB to complete manually, so I'm afraid it doesn't get "dog-fooded" by me. An ideal fix would also include updating racket-tests.el to package-install company-mode, enable it, and try a few simple things. Maybe I should add that as an issue, in case anyone else has time to help by adding that. (I've been kind of swamped by paid contract work lately.)

[greghendershott]

I can't seem to reproduce this. When I type de, it doesn't try to connect. Instead I get a company-mode window showing candidates. I can press RET to choose one and continue. C-u ' takes me back to the org file, as expected.

racket-mode needn't start a racket process and the REPL, to get completion candidates. Ever since commit b0296d9 in July 2018, if the REPL isn't running, it doesn't try to start it. Instead it simply uses for completion candidates the same default list used for font-lock. The idea was to avoid exactly this kind of problem trying to start Racket while you're typing.

Your details suggest you have quite recent racket-mode, including this commit:

(racket--el-source-dir "/home/bhrgunatha/.emacs.d/.cask/26.1/elpa/racket-mode-20181206.329/")
(racket--rkt-source-dir "/home/bhrgunatha/.emacs.d/.cask/26.1/elpa/racket-mode-20181206.329/racket/")

So I am super confused why you're seeing this. 🙁

[bhrgunatha]

I'm not sure why the REPL would get started but I it seems likely that is where the issue lies.
After the window & buffer to edit source code from org mode appears:

Type (
Type d
Type e

emacs hangs - 1 core is 100% CPU in process monitor

CTRL-g - emacs returns to normal
run.rkt appears in process monitor as a separate process

Here's what appears in the *Messages* buffer during this experiment::

Indenting region...done
Checking Racket version ...
Starting racket to run /home/bhrgunatha/.emacs.d/.cask/26.1/elpa/racket-mode-20181206.329/racket/run.rkt ...
Still trying to connect to racket-command process on port 55555 ... [231391 times]
Connected to racket-command process on port 55555 after 2 attempt(s)

Seems like a tight loop checking for the racket command to appear is causing the CPU spike. CTRL-g ends that for some reason - racket command process appears and things come back to normal.

I'm not sure how or why company-mode is related but if I disable company-mode there is no hang. Re-enabling indeed cusses the hang to return. My company version is company-20190109.2336 (again from melpa).

I wonder if it's related to emacs 26.1's new threading behaviour?

EDIT: Further: it doesn't appear to be the company-mode/racket-mode combination - that works well for me - it seems to be org-mode/company-mode/racket-mode.

[bhrgunatha]

After tinkering around, trying to track down the cause issue I've edited several files in racket-mode, org-mode and company-mode.
I've just nuked my whole package folder and re-installed all packages and now I can't reproduce the issue.

The README does say it might be necessary to delete and re-install racket-mode for updates and I think that has fixed my problem.

I really should have done that first, so sorry to have wasted your time.

[greghendershott]

No worries. At the time you reported this, #344 was still open so it seemed credible you'd found a variant (plus I was already looking at #344 so it wasn't much incremental effort to look at yours).