Original Redmine Comment
Author Name: Slava Yuzhaninov
Original Date: 2016-11-05T16:20:14Z

+After some investigation:+
It happens because these keywords are defined as part of @verilog-mode-abbrev-table@. I have a globally set @abbrev-mode@, thus verilog-mode inherits this variable and automatically evokes the corresponding skeletons. In my opinion, the summoning of the skeletons should be conditional, i.e. not part of a comment.

Tried to do something like that:

(verilog-define-abbrev verilog-mode-abbrev-table "if" "" (unless (inside-comment-q) 'verilog-sk-if))
</code>

Where @inside-comment-q@ is (seems to work fine):

(defun inside-comment-q ()
  (interactive)
  (let ((result (nth 4 (syntax-ppss))))
     (message "%s" result)
     result))
</code>

Didn't work... It's still evoking the skeleton.

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2016-11-14T17:59:27Z

I can see how this would be useful. Alas the skeletons use the standard Emacs packages, rather than code in verilog-mode itself. I looked at a few modes and none seem to stop expansion on comments, nor could I find an easy way to provide this; all I could find was one forum post which showed a very complicated way which wouldn't be maintainable. If you find a way to cleanly solve this please let us know and we'll look at adding the code to verilog-mode. Sorry.

Original Redmine Comment
Author Name: Slava Yuzhaninov
Original Date: 2016-12-24T19:42:32Z

Wilson Snyder wrote:

I can see how this would be useful. Alas the skeletons use the standard Emacs packages, rather than code in verilog-mode itself. I looked at a few modes and none seem to stop expansion on comments, nor could I find an easy way to provide this; all I could find was one forum post which showed a very complicated way which wouldn't be maintainable. If you find a way to cleanly solve this please let us know and we'll look at adding the code to verilog-mode. Sorry.

Hi.

I had some time to play a bit with the conditional skeleton expansion and I think that there is a reasonble solution:

1. I modified the verilog-define-abbrev entry for example:
   ```(verilog-define-abbrev verilog-mode-abbrev-table "if" "" `context-abbrev-verilog-sk-if)

Instead of the former:
```<code class="clojure">(verilog-define-abbrev verilog-mode-abbrev-table "if" "" `verilog-sk-if)</code>

2. Defined a contextual intermediate function context-abbrev-verilog-sk-if:

(defun context-abbrev-verilog-sk-if ()
  (if (or (inside-comment-q) (inside-string-q)) (unexpand-abbrev) (verilog-sk-if)))
</code>
``` which rolls back the abbreviation expansion if the position is inside a comment or inside a string, otherwise the expansion skeleton is called as before.



3. The *inside-comment-q* and *inside-string-q* functions are (referenced from @http://ergoemacs.org/emacs/elisp_determine_cursor_inside_string_or_comment.html@):

```<code class="clojure">
(defun inside-comment-q ()
  "Returns non-nil if inside comment, else nil. Result depends on syntax table's comment CHAR."
  (interactive)
  (let ((result (nth 4 (syntax-ppss)))) result))

(defun inside-string-q ()
  "Returns non-nil if inside string, else nil. Result depends syntax table's string quote CHAR."
  (interactive)
  (let ((result (nth 3 (syntax-ppss)))) result))
</code>

4. The solution is not 100%, because the abbreviation rolling back disrupts the cursor position. Thus additional treatment to restore the old position is required. I guess that the LISP's dedicated save-excursion function could be used, but I'm not so great in LISP.

What do you think?
Thanks...

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2017-01-05T00:11:00Z

Seems a reasonable track.

Functions in verilog-mode have to start with verilog, so e.g. instead of context-abbrev-verilog-sk-if use verilog-context-abbrev-sk-if

You can try using verilog-inside-comment-or-string-p instead of the other functions, which should not move the point.

Please also make a new defvar, e.g. verilog-sk-inside-comments which is used inside your function to select if the function simply behaves as before (t), or checks for comments (nil). That way if something breaks it can be easily disabled.

If you get it all working, please put it into a patch file and I'll review and we can get it merged upstream, thanks for experimenting.

Original Redmine Comment
Author Name: Slava Yuzhaninov
Original Date: 2017-01-28T23:22:32Z

Hi.

Here is my "primitive" version of the skeleton wrapper function, which works fine:

(defun verilog-context-abbrev-sk-if ()
  "Wrapper function for conditional `if keyword abreviation
  expansion, which ignored when the cursor is inside a
  comment/string."
  ;;  Save the last expanded abbreviation text, before nullified by the 'unexpand-abbrev function:
  (setq last-expand-abbrev-text-backup (symbol-value 'last-abbrev-text))
  (if (verilog-in-comment-or-string-p)
       ;; Inside a comment/string ==> Unreal source-code:
       (progn
         (unexpand-abbrev)                                       ;Roll back the abbreviation expansion
         (forward-char (length last-expand-abbrev-text-backup))) ;Correct cursor position
     ;; Real source-code:
     (verilog-sk-if)))
</code>

Few points:

1. I used the verilog-mode built-in @verilog-in-comment-or-string-p@ function rather than new custom functions.
2. All the functions (for each keyword) look almost the same, so maybe the wrapper function might be generalized to get the skeleton as an input argument.
3. The relevant section of the modified code is attached as a verilog-context-abbrev-sk-patch.el file.

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2017-02-09T22:18:12Z

If you'd like your patch merged in, could you please make your patch relative to the latest verilog-mode.el (from git or download here) and I'll try to merge it in.

Original Redmine Comment
Author Name: Slava Yuzhaninov
Original Date: 2017-02-10T17:40:11Z

Wilson Snyder wrote:

please make your patch relative to the latest verilog-mode.el (from git or download here) and I'll try to merge it in.

Here is the latest verilog-mode.el with the aforementioned additions

Original Redmine Comment
Author Name: Wilson Snyder (@wsnyder)
Original Date: 2017-02-12T23:28:53Z

Fixed in git and verilog-mode-2017-02-12-224d7a2-vpo

Attached is the patch I cleaned up if it's historically valuable. But then I found out there was a global way to suppress expansion starting in Emacs 23.1 which is what it now calls. Sorry I didn't know about this earlier.