Re: Src blocks fontification issue

[Dan Davison <davison@stats.ox.ac.uk>]

Julien Fantin <julien.fantin@gmail.com> writes:

> I tried opening the file with a bare emacs and the problem persists.
>
> I use true indentation in my files, which I think isn't what most
> people do, so maybe the parser gets confused by a combination of
> blocks not starting at column 0, and some other formatting, but I'm
> failing to see the pattern...

Me too. While you may well be right that I have broken something, first
let's concentrate on what appears to be a long-standing bug; in fact I
think it has never worked correctly. Basic fontification of src blocks
elements was introduced over a year ago at this commit

--8<---------------cut here---------------start------------->8---
commit 476493d3b300b8dc379ea7e42c974461d6f3dfb5
Author: Carsten Dominik <carsten.dominik@gmail.com>
Date:   Sat May 23 22:58:38 2009 +0200

    Fontification: Better treatment of #+ lines and blocks
    
    The content of blocks like #+begin_example will be marked with a
    separate face.  That same face is also used for single lines
    introducing text for specific backends.
--8<---------------cut here---------------end--------------->8---

Priot to that commit the whole block, and the begin_src end_src lines
had no special fontification. After that commit, a test file (below)
based on yours is not correctly fontified on startup. It is correctly
fontified when font-lock is triggered subsequently by some keyborad
input nearby. But it seems very sensitive to changes in the test file
and I haven't worked out what's going on.

Test file: the final code block is not fontified, nor are the BEGIN_SRC
and END_SRC lines. (emacs 24, org commit 476493d3b300 and subsequent)

--8<---------------cut here---------------start------------->8---
* Setup
  #+BEGIN_SRC emacs-lisp
    (add-to-list 'auto-mode-alist '("\\.\\(org\\|org_archive\\|txt\\)$" . org-mode))
    (add-hook 'org-mode-hook
              '(lambda ()
                 (auto-fill-mode t)
                 (define-key org-mode-map (kbd "RET") 'org-return-indent)
                 (flyspell-mode 1)))
    
    
    ;; windmove fix
    (add-hook 'org-shiftup-final-hook 'windmove-up)
    (add-hook 'org-shiftleft-final-hook 'windmove-left)
    (add-hook 'org-shiftdown-final-hook 'windmove-down)
    (add-hook 'org-shiftright-final-hook 'windmove-right)
  #+END_SRC
** Helpers
  #+BEGIN_SRC emacs-lisp
   (defun is-bogus-file-p (file)
     "Return t if file is a temporary file as in .#foo or a . .. reference returned by directory-files"
     (or (member file '("." ".."))
         (string-match "^.#.+" file)))
  #+END_SRC
* Customization
  #+BEGIN_SRC emacs-lisp
  (setq org-return-follows-link t
        org-special-ctrl-a/e t
        org-special-ctrl-k t
        org-completion-use-ido t
        org-log-into-drawer t
        org-startup-indented nil
        org-hide-leading-stars t
        org-startup-folded t
        org-empty-line-terminates-plain-lists t
        org-adapt-indentation t
        org-yank-folded-subtrees t
        org-M-RET-may-split-line nil
        org-enforce-todo-dependencies t
        org-enforce-todo-checkbox-dependencies t)

  ;; (setq org-blank-before-new-entry '((heading . always)
  ;;                                    (plain-list-item . always)))
  #+END_SRC
--8<---------------cut here---------------end--------------->8---


Dan

>
> Here is one of the incriminated files:
>
> http://pastebin.com/raw.php?i=yxdUmpKx
>
> Hope this is any help...
>
> Cheers
>
> On Fri, Sep 3, 2010 at 12:48 AM, Dan Davison <davison@stats.ox.ac.uk> wrote:
>> Julien Fantin <julien.fantin@gmail.com> writes:
>>
>>> First congratulations on adding in-line fontification to source
>>> blocks, that's a great addition.
>>>
>>> However, since I pulled those changes, a recurring issue I've been
>>> having has become even more annoying.
>>>
>>> When opening an org file, some src blocks headers don't seem to get
>>> parsed properly, and now as a result, don't get fontified.
>>> They are functional though, since they properly get tangled, and C-c '
>>> performs as expected.
>>> If I edit the header line, or the ones directly before and after it,
>>> the block gets back to normal.
>>>
>>> Here is a screenshot to make things clearer, as you can see there are
>>> no differences in formatting, and those properties drawer don't have
>>> anything weird, just a single ID.
>>>
>>> http://imagebin.org/112413
>>>
>>> I was meaning to post a screenshot with whitespace-mode on to stress
>>> that there are no apparent formatting differences, but it turned out
>>> activating it fixed the issue !
>>>
>>> Not sure what I can do to further help debugging this ?
>>
>> Hi Julien,
>>
>> I don't have any clever ideas about this, so I'm just going to suggest
>> the obvious: first a sanity check that it doesn't occur with emacs
>> -Q. Then some sort of quasi-binary search through your emacs
>> customisations. It looks like you're storing your emacs customisation in
>> an Org file, so you can knock out entire trees with :tangle: no
>> properties.
>>
>> Dan
>>
>> p.s. On a completely unrelated note, if you haven't already tried it:
>>
>> (add-hook 'emacs-lisp-mode-hook 'pretty-lambdas)
>>
>> where the `pretty-lambdas' function is in starter-kit-defuns.org at
>> git://github.com/eschulte/emacs-starter-kit.git
>>
>> (defun pretty-lambdas ()
>>  (font-lock-add-keywords
>>   nil `(("(?\\(lambda\\>\\)"
>>          (0 (progn (compose-region (match-beginning 1) (match-end 1)
>>                                    ,(make-char 'greek-iso8859-7 107))
>>                    nil))))))
>>
>>>
>>> Cheers
>>>
>>> _______________________________________________
>>> Emacs-orgmode mailing list
>>> Please use `Reply All' to send replies to the list.
>>> Emacs-orgmode@gnu.org
>>> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
>>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode