Export options being ignored

Export options being ignored

[Eden Cardim]

,----[ test.org ]
| #+OPTIONS: <:nil
| ** TODO Estimate effort for all of today's tasks
|    SCHEDULED: <2011-05-05 Thu 06:40 ++1d>
`----

Exporting the above file to pdf results in a file containing the
SCHEDULED timestamp in it.

I'm running org from git master, commit
cd0446243861487096983f09e545105551af7f10

Apparently, all the symbol settings (^:t |:t etc...) exhibit the same
behaviour. Options like toc:t, num:t, etc. do work. I'm not familiar
with the code that does the parsing of the options so I have no idea
where to look in order to debug this. Setting org-export-with-timestamps
etc. does work as documented, so I'd say something is going wrong with
the parsing.

-- 
Eden Cardim
Software Engineer
edencardim.com
+55 73 9986-3963

Re: Export options being ignored

[Nick Dokos]

Eden Cardim <edencardim@gmail.com> wrote:

> 
> ,----[ test.org ]
> | #+OPTIONS: <:nil
> | ** TODO Estimate effort for all of today's tasks
> |    SCHEDULED: <2011-05-05 Thu 06:40 ++1d>
> `----
> 
> Exporting the above file to pdf results in a file containing the
> SCHEDULED timestamp in it.
> 
> I'm running org from git master, commit
> cd0446243861487096983f09e545105551af7f10
> 
> Apparently, all the symbol settings (^:t |:t etc...) exhibit the same
> behaviour. Options like toc:t, num:t, etc. do work. I'm not familiar
> with the code that does the parsing of the options so I have no idea
> where to look in order to debug this. Setting org-export-with-timestamps
> etc. does work as documented, so I'd say something is going wrong with
> the parsing.
> 

Reproduced and git bisected:

,----
| aa6dba8a74016587755c250bb8cc4743a4082ea1 is the first bad commit
| commit aa6dba8a74016587755c250bb8cc4743a4082ea1
| Author: Lawrence Mitchell <wence@gmx.li>
| Date:   Thu Jan 20 18:23:22 2011 +0000
| 
|     Only match complete words in org-export-add-options-to-plist
|     
|     * org-exp.el (org-export-add-options-to-plist): Require match to start
|     at a word-boundary.
|     
|     Previously, if an option was the suffix of another option (such as TeX
|     and LaTeX) the setting for the former would propagator to the latter.
|     This seems like an unintended consequence of a lax regexp in
|     org-export-add-options-to-plist.  This patch allows options to share a
|     suffix with another option by requiring that the match against an
|     option starts at a word-boundary.
| 
| :040000 040000 44d426100f5c092112e149a891eb2e043a8ca404 c76627facb6988aa08d4d1fa9f1ab2d9b4aa2f45 M	lisp
`----

Lawrence, can you take a look?

Thanks,
Nick

Re: Export options being ignored

[Nick Dokos]

Nick Dokos <nicholas.dokos@hp.com> wrote:

> ,----
> | aa6dba8a74016587755c250bb8cc4743a4082ea1 is the first bad commit
> `----
> 

Taking a look at the commit:

,----
| commit aa6dba8a74016587755c250bb8cc4743a4082ea1
| Author: Lawrence Mitchell <wence@gmx.li>
| Date:   Thu Jan 20 18:23:22 2011 +0000
| 
|     Only match complete words in org-export-add-options-to-plist
|     
|     * org-exp.el (org-export-add-options-to-plist): Require match to start
|     at a word-boundary.
|     
|     Previously, if an option was the suffix of another option (such as TeX
|     and LaTeX) the setting for the former would propagator to the latter.
|     This seems like an unintended consequence of a lax regexp in
|     org-export-add-options-to-plist.  This patch allows options to share a
|     suffix with another option by requiring that the match against an
|     option starts at a word-boundary.
| 
| diff --git a/lisp/org-exp.el b/lisp/org-exp.el
| index a265c3b..4a10303 100644
| --- a/lisp/org-exp.el
| +++ b/lisp/org-exp.el
| @@ -830,7 +830,7 @@ security risks."
|        (let ((op org-export-plist-vars))
|  	(while (setq o (pop op))
|  	  (if (and (nth 1 o)
| -		   (string-match (concat (regexp-quote (nth 1 o))
| +		   (string-match (concat "\\<" (regexp-quote (nth 1 o))
|  					 ":\\([^ \t\n\r;,.]*\\)")
|  				 options))
|  	      (setq p (plist-put p (car o)
`----

explains the problem: \< matches the empty string at the beginning of a
word (i.e. if the syntax class of the next character is "word") but it
does not at the beginning of a char that is of some other syntax class
(I think it will not match anything in this case).  So Eden diagnosed it
correctly: it *is* a parsing problem and it *does* involve the non-word
options.

At this point, the cure looks worse than the disease, so this commit should
probably be reverted.

Thanks,
Nick

Re: Export options being ignored

[Carsten Dominik]

On 5.5.2011, at 07:08, Nick Dokos wrote:

> Nick Dokos <nicholas.dokos@hp.com> wrote:
> 
>> ,----
>> | aa6dba8a74016587755c250bb8cc4743a4082ea1 is the first bad commit
>> `----
>> 
> 
> Taking a look at the commit:
> 
> ,----
> | commit aa6dba8a74016587755c250bb8cc4743a4082ea1
> | Author: Lawrence Mitchell <wence@gmx.li>
> | Date:   Thu Jan 20 18:23:22 2011 +0000
> | 
> |     Only match complete words in org-export-add-options-to-plist
> |     
> |     * org-exp.el (org-export-add-options-to-plist): Require match to start
> |     at a word-boundary.
> |     
> |     Previously, if an option was the suffix of another option (such as TeX
> |     and LaTeX) the setting for the former would propagator to the latter.
> |     This seems like an unintended consequence of a lax regexp in
> |     org-export-add-options-to-plist.  This patch allows options to share a
> |     suffix with another option by requiring that the match against an
> |     option starts at a word-boundary.
> | 
> | diff --git a/lisp/org-exp.el b/lisp/org-exp.el
> | index a265c3b..4a10303 100644
> | --- a/lisp/org-exp.el
> | +++ b/lisp/org-exp.el
> | @@ -830,7 +830,7 @@ security risks."
> |        (let ((op org-export-plist-vars))
> |  	(while (setq o (pop op))
> |  	  (if (and (nth 1 o)
> | -		   (string-match (concat (regexp-quote (nth 1 o))
> | +		   (string-match (concat "\\<" (regexp-quote (nth 1 o))
> |  					 ":\\([^ \t\n\r;,.]*\\)")
> |  				 options))
> |  	      (setq p (plist-put p (car o)
> `----
> 
> explains the problem: \< matches the empty string at the beginning of a
> word (i.e. if the syntax class of the next character is "word") but it
> does not at the beginning of a char that is of some other syntax class
> (I think it will not match anything in this case).  So Eden diagnosed it
> correctly: it *is* a parsing problem and it *does* involve the non-word
> options.
> 
> At this point, the cure looks worse than the disease, so this commit should
> probably be reverted.

This is fixed now, by looking for white space instead of beginning-of-word.
Thanks for the analysis.

- Carsten

Re: Export options being ignored

[Lawrence Mitchell]

Carsten Dominik wrote:
> On 5.5.2011, at 07:08, Nick Dokos wrote:

[...]

>> explains the problem: \< matches the empty string at the beginning of a
>> word (i.e. if the syntax class of the next character is "word") but it
>> does not at the beginning of a char that is of some other syntax class
>> (I think it will not match anything in this case).  So Eden diagnosed it
>> correctly: it *is* a parsing problem and it *does* involve the non-word
>> options.

> This is fixed now, by looking for white space instead of beginning-of-word.
> Thanks for the analysis.

Indeed, thanks.

Lawrence
-- 
Lawrence Mitchell <wence@gmx.li>

Re: Export options being ignored

[Eden Cardim]

>>>>> "Carsten" == Carsten Dominik <carsten.dominik@gmail.com> writes:

    Carsten> This is fixed now, by looking for white space instead of
    Carsten> beginning-of-word. Thanks for the analysis.

Still needs the following patch for it to work:

--8<---------------cut here---------------start------------->8---
diff --git a/lisp/org-exp.el b/lisp/org-exp.el
index 5bf1d1b..cda1f98 100644
--- a/lisp/org-exp.el
+++ b/lisp/org-exp.el
@@ -864,7 +864,7 @@ security risks."
                                 options))
              (setq p (plist-put p (car o)
                                 (car (read-from-string
-                                      (match-string 1 options))))))))))
+                                      (match-string 2 options))))))))))
   p)
 
 (defun org-export-add-subtree-options (p pos)
--8<---------------cut here---------------end--------------->8---

-- 
Eden Cardim
Software Engineer
edencardim.com
+55 73 9986-3963

Re: Export options being ignored

[Carsten Dominik]

On 5.5.2011, at 12:40, Eden Cardim wrote:

>>>>>> "Carsten" == Carsten Dominik <carsten.dominik@gmail.com> writes:
> 
>    Carsten> This is fixed now, by looking for white space instead of
>    Carsten> beginning-of-word. Thanks for the analysis.
> 
> Still needs the following patch for it to work:

Yes, thanks.

- Carsten

> 
> --8<---------------cut here---------------start------------->8---
> diff --git a/lisp/org-exp.el b/lisp/org-exp.el
> index 5bf1d1b..cda1f98 100644
> --- a/lisp/org-exp.el
> +++ b/lisp/org-exp.el
> @@ -864,7 +864,7 @@ security risks."
>                                 options))
>              (setq p (plist-put p (car o)
>                                 (car (read-from-string
> -                                      (match-string 1 options))))))))))
> +                                      (match-string 2 options))))))))))
>   p)
> 
> (defun org-export-add-subtree-options (p pos)
> --8<---------------cut here---------------end--------------->8---
> 
> -- 
> Eden Cardim
> Software Engineer
> edencardim.com
> +55 73 9986-3963
> 
>