Re: [Boost-docs] Doxygen (non)-processing of namespace detail

Subject: Re: [Boost-docs] Doxygen (non)-processing of namespace detail
From: Daniel James (dnljms_at_[hidden])
Date: 2016-03-10 19:56:09

On 9 March 2016 at 17:17, Paul A. Bristow <pbristow_at_[hidden]> wrote:
>
>   <!-- The namespace used to identify code that should not be  processed at all. -->
>   <xsl:param name="boost.doxygen.detailns">detail</xsl:param>
>
>   <!-- The substring used to identify unspecified types that we can't
>        mask from within Doxygen. This is a hack (big surprise). -->
>   <xsl:param name="boost.doxygen.detail"><xsl:value-of select="$boost.doxygen.detailns"/>::</xsl:param>

As these are parameters, you can specify the value in your jamfile. So
if you don't want it, set boost.doxygen.detailns to a string that will
never match, e.g.
<xsl:param>boost.doxygen.detailns=dont-ignore-detail-namespace

> My experiments seem to show that namespace 'details' is also not processes (but detai is).

Looking at the xsl it uses 'contains' to check for detail namespaces,
so it's going to match anything including 'detail'. It could be
changed to something like: contains(concat('::',$fullname,'::'),
concat('::',$boost.doxygen.detailns,'::')). Although I wouldn't be
surprised if some documentation relies on the existing behaviour.

> Since one might want Doxygen to process details for maintenance, this seems a rather blunt mechanism, compared to
>
> 1 Not processing header files with Doxygen that contain detail sections by choice of input in doxyfile/ jamfile.

Can't do that in XSL, you'd have to work out how to do it in boost
build. It might just be a case of using globs to select the right
files.

> 2 Using the Doxygen mechanism \cond BOOST_DETAIL ... \endcond to control what is 'Doxygenated' into the boostbook.

Maybe.