Re: Possible buglet in latex export

[Robert Goldman <rpgoldman@sift.info>]

Tim Burt wrote:
> Robert Goldman writes:
>  > When I do a latex export, a simple URL in text, or a simple link url of
>  > the form [[URL]] --- with no description --- gets emitted as
>  > 
>  > \href{URL}{URL}
>  > 
>  > which causes Latex to crash for me.
> The first guess is that the string 'URL' in the post is not the actual
> string that "causes Latex to crash".  The next guess is that 'URL' is
> actually a string with one of LaTeX's 10 special characters, with a
> popular one being underscore (_) in this context.
> 
>  > 
>  > I believe that this is because it won't accept a URL as the second
>  > argument to href.
> The second argument to \href is simply a string that LaTeX can render,
> so a URL is fine.
> 
>  > 
>  > Changing the \href command to \url fixes the latex problem.
>  > 
>  > So I wonder if we need to catch this special case of a description-less
>  > URL and treat it specially in latex export for the benefit of latex's
>  > hyperref package.
>  > 
>  > I'm not an expert on hyperref, by any means, nor do I know the innards
>  > of latex export, so I could be missing something here.  However the
>  > description of \url in the hyperref manual says the following:
>  > 
>  >   \url{URL}
>  > 
>  > Similar to \href{URL}{\nolinkurl{URL}}
>  > 
>  > ...which suggests to me that using the URL without some kind of magical
>  > protection (provided by \nolinkurl) may lead to bad things.
> A URL can be used, but any special characters must be protected
> somehow.  The \nolinkurl in the second argument provides such
> protection, as does the \url solution mentioned above.
> 

Thanks.  This suggests that replacing \href{URL}{URL} with
\href{URL}{\nolinkurl{URL}} might be a good strategy for Latex export.

I am not exactly sure why the un-protected URL causes problems for me
and not for Nick.  However, I am trying to generate latex for beamer,
which seems very tricky, and perhaps that's what's going wrong here ---
I'm getting beamer's internal state messed up.  I'm not sure; will
report if I can disentangle it.

Thanks for the help, both of you.

best,
R