Re: [PATCH] find-file-noselect-1

[Nick Roberts]

 > I suggest you change the progn to condition-case so as to catch any
 > errors that happen in gdb-enqueue-input.  And if it does ever catch an
 > error, it could set gud-comint-buffer to nil so it won't do anything
 > on future calls to find-file.  That should make it safe.

(defun gdb-find-file-hook ()
  (if (and (boundp 'gud-comint-buffer)
           ;; in case gud or gdb-ui is just loaded.
           gud-comint-buffer
           (buffer-name gud-comint-buffer)
           (with-current-buffer gud-comint-buffer
             (eq gud-minor-mode 'gdba)))
      (condition-case nil
        (gdb-enqueue-input
         (list (concat "list " (file-name-nondirectory buffer-file-name)
                       ":1\n")
               `(lambda () (gdb-set-gud-minor-mode ,buffer-file-name)))))
    (error nil)))

Is this what you mean? I don't think that gud-comint-buffer should be set to
nil because that would prevent the GDB session from recovering.

 > 
 >     + (defun gdb-set-gud-minor-mode (file)
 >     +   "Set gud-minor-mode from find-file if appropriate."
 >     +   (goto-char (point-min))
 >     +   (unless (search-forward "No source file named " nil t)
 >     +       (with-current-buffer
 >     +          (find-file-noselect file)
 > 
 > Why call find-file-noselect there?  If this is meant to operate on the
 > file that was just visited, it already has a buffer, and it is the
 > current buffer when gdb-find-file-hook runs.  Why not just use
 > that buffer?

That might have been true but I'm now using this function to address Kim's
point about enabling gud-minor-mode for existing buffers.

 > +    (save-current-buffer
 > 
 > What's the reason for that?  The contents don't change buffers.

This was just a legacy from earlier code. I've removed it.

Since the changes are all in gdb-ui.el, I feel more comfortable installing
them. However, having asked for people's opinion, I'll wait for some approval
before doing this.

Nick