Hi Richard,

I do really appreciate your taking time to further explain the use of
CUSTOM_ID.

Now I can understand why this CUSTOM_ID-based solution is better and, in
some cases as in your example, needed.

With version 7.x, what I did was to reference sections in main texts (not
in embedded LaTeX blocks) using 'C-c l' on a sections header and 'C-c C-l'
in a part referring to it (i.e., no manual setting & typing involved).

Fortunately, for most journal & conference papers in my area (i.e., up to
~15 pages), there is virtually no chance that more than one sections have
the same title at the same level; note that "*Numerical Examples" and
"**Numerical Examples" are different. This may be the case for a thesis as
well, much longer though.

Except for the said case of embedded LaTeX blocks, therefore, the use of
CUSTOM_ID with manual labelling is something I want to avoid if possible.

By the way, as said, [[*Foo][Section Foo]] does not produce the link in
LaTeX export any longer; version 8.x generates just "\texttt{Section Foo}"
unlike version 7.x.

But I do agree that new export engines are more flexible and powerful
(e.g., filters) and will see whether and how all these compatibility issues
will be addressed; otherwise, I would have to update tens of papers based
on version 7.x org-mode, which is my major platform/format for academic
publications and teaching materials.

Again, many thanks for your sharing valuable experience and great insight
on these!
Joseph

On Fri, Sep 19, 2014 at 11:49 PM, Richard Lawrence <
richard.lawrence@berkeley.edu> wrote:

> Hi Joseph,
>
> "Kyeong Soo (Joseph) Kim" <kyeongsoo.kim@gmail.com> writes:
>
> > Great thanks for your answering my two questions, especially sharing your
> > experience for cross-referencing in LaTeX export.
> >
> > Though your suggestion for cross-referencing is an excellent workaround,
> I
> > wonder whether there is any benefit using org-mode in writing papers in
> > this case; for instance, manual setting of CUSTOM_ID would be a hassle,
> > considering that it can be easily & automatically done in LaTeX with
> > AUCTeX/RefTeX's label function.
>
> Don't you need to make sure that your sections have a consistent label
> in order to refer to them?
>
> At any rate, one of the reasons I use this solution is that I want to be
> able to easily use plain LaTeX at points where I discover that I can't
> make Org do everything I want.  One of the cases I was worried about,
> for example, is having a section reference inside an embedded LaTeX
> block.  If you don't know the label ahead of time, you can't use a raw
> \ref command inside #+BEGIN_LATEX...#+END_LATEX.  So I like being able
> to set the labels via CUSTOM_ID on an as-needed basis. (I don't do it
> for every section.)
>
> You might be able to get away with less; it looks to me like the
> [[*Foo][Section Foo]] still produces a link to a headline in LaTeX
> output.  Does that not work for you?
>
> > I still don't know why the new export engine has brought so many
> > compatibility issues in its behavior.  It seems that I'd better go
> > back to the prior version (7.X) in order to focus on my research.
>
> The new export engine was a complete re-write, and came with an attempt
> to define and standardize Org syntax.  Some of the incompatibilities are
> due to that standardization: the old LaTeX and HTML exporters might
> treat some text differently, for example, or not deal with some corner
> case, so in writing a new export engine, it was necessary to define the
> correct behavior.  Other incompatibilities are just bugs that come with
> any new piece of software; the exporter is under heavy development and
> those are constantly getting fixed.
>
> The new exporter really is a *lot* better, and in my experience it is
> worth the effort to upgrade.  I found myself bumping into the
> limitations of the pre-8.0 exporter very frequently.  If you don't, you
> can wait, but I think you'll find the new exporter a lot easier to work
> with and to get help with.
>
> Best,
> Richard
>
> OpenPGP Key ID: CF6FA646
> Fingerprint: 9969 43E1 CF6F A646
>
> (See http://www.ocf.berkeley.edu/~rwl/encryption.html for more
> information.)
>