Re: [O] [RFC] [PATCH] ox-latex: support :float no with caption for minted listings

[Aaron Ecay]

Hi Nicolas,

2014ko abuztuak 23an, Nicolas Goaziou-ek idatzi zuen:
> 
> Aaron Ecay <address@hidden> writes:
> 
>> It’s not adding any new functionality to the exporter,
> 
> Well, it is, since "captionof" command doesn't exist in Org default
> packages.

I guess this was unclear.  What I meant was that the patch is not
introducing any new expressivity on the org-mode side.  There is a
situation (described in my previous mail) that is a valid combination of
org configuration variables and buffer text, which produces inconsistent
latex output.  This patch fixes that.

> 
>> but rather covering one particular combination of already-existing
>> options (caption provided, :float nil, org-latex-listings = minted)
>> that does not produce sensible output presently. Should I fix the
>> nil/no issue and send a new patch?
> 
> With your patch, when encountering the combination above, the export
> back-end will introduce a "captionof" command, which requires loading
> "caption" package with a specific option (i.e., compatibility=false).

Why is the compatibility=false option needed?  I can’t figure this
out.

> 
> If this is explained nowhere, it is hardly a fix. If you document it,
> you introduce support for "caption" package in Org. Is there a strong
> reason for that?

“Pretty” source code export in all cases requires adding certain packages
to the default, as explained in the docstring of ‘org-latex-listings’.
The patch adds discussion of the caption requirement there, so there is
no “if” about the documentation.

I don’t understand what you’re saying about introducing “support” for
the caption package.  The patch uses one specific feature of the caption
package in one specific place to accomplish one specific goal.  There
are no other changes needed anywhere else in the codebase than these
couple of lines in one function (plus the documentation).

The motivating reason is described in my first paragraph above – making
the output consistent for all combinations of options.

Thanks,

-- 
Aaron Ecay