[libcamera-devel] [PATCH 4/5] libcamera: Document private members

[Laurent Pinchart]

Hi Jacopo,

Thank you for the patch.

On Thursday, 3 January 2019 19:38:58 EET Jacopo Mondi wrote:
> The newly added MediaDevice::setLink() function is defined as private,
> but it is worth being documented, as it is called from the friend method
> MediaLink::setup().
> 
> In the library code base, several private methods and fields are
> documented, but do not show up in the generated documentation output.
> Change doxygen settings to output processed documentation for the
> private fields and methods, when proper doxygen comments are applied to
> them (and do not complain if private members are not documented at all).
> 
> Signed-off-by: Jacopo Mondi <jacopo at jmondi.org>
> ---
>  Documentation/Doxyfile.in      |  4 ++--
>  src/libcamera/media_device.cpp | 22 ++++++++++++++++++++++
>  2 files changed, 24 insertions(+), 2 deletions(-)
> 
> diff --git a/Documentation/Doxyfile.in b/Documentation/Doxyfile.in
> index b1a70d3..16dcccd 100644
> --- a/Documentation/Doxyfile.in
> +++ b/Documentation/Doxyfile.in
> @@ -442,7 +442,7 @@ EXTRACT_ALL            = NO
>  # be included in the documentation.
>  # The default value is: NO.
> 
> -EXTRACT_PRIVATE        = NO
> +EXTRACT_PRIVATE        = YES
> 
>  # If the EXTRACT_PACKAGE tag is set to YES, all members with package or
> internal # scope will be included in the documentation.
> @@ -487,7 +487,7 @@ EXTRACT_ANON_NSPACES   = NO
>  # section is generated. This option has no effect if EXTRACT_ALL is
> enabled. # The default value is: NO.
> 
> -HIDE_UNDOC_MEMBERS     = NO
> +HIDE_UNDOC_MEMBERS     = YES
> 
>  # If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
>  # undocumented classes that are normally visible in the class hierarchy. If
> set diff --git a/src/libcamera/media_device.cpp
> b/src/libcamera/media_device.cpp index b86d0c4..48aa805 100644
> --- a/src/libcamera/media_device.cpp
> +++ b/src/libcamera/media_device.cpp
> @@ -641,6 +641,28 @@ bool MediaDevice::populateLinks(const struct
> media_v2_topology &topology) return true;
>  }
> 
> +/**
> + * \brief Apply \a flags to a link between two pads
> + * \param source The source pad
> + * \param sink The sink pad

How about passing the link pointer instead ?

> + * \param flags The link flags
> + *
> + * Implements 'raw' link handling, as this functions applies \a flags,
> + * (whose only accepted values are the ones defined by the Media Controller
> + * APIs in MEDIA_LNK_FL_* macros) to a link between two pads. No
> correctness + * checks on the link existence and validity of \a flags is
> performed. The + * function assumes the \a source and \a sink pads are
> connected, and the + * supplied \a flags applies to the link (ie. immutable
> links cannot be + * disabled).

"This function applies the link \a flags (as defined by the MEDIA_LNK_FL_* 
macros from the Media Controller API) to the given \a link. It implements low-
level link setup as it performs no checks on the validity of the \a flags, and 
assumes that the supplied \a flags are valid for the link (e.g. immutable 
links cannot be disabled)."

> + *
> + * This function wraps the raw MEDIA_IOC_SETUP_LINK ioctl, and shouldn't be
> + * called directly by any other class or method but the
> + * MediaLink::setup(bool enable) one, declared as friend for this reason.
> + *
> + * \sa MediaLink::setup(bool enable)
> + *
> + * \return 0 for success, negative error number otherwise
> + */
>  int MediaDevice::setLink(const MediaPad *source, const MediaPad *sink,
>  			 unsigned int flags)

Should this be named setupLink() to refer to the MediaController terminology ?

>  {

-- 
Regards,

Laurent Pinchart