Hi Nicolas,
> > > > > +The structure passed for each FFT operation is given below, with
> > > > > +the operation flags forming a bitmask in the ``op_flags`` field.
> > > > > +
> > > > > +.. code-block:: c
> > > > > +
> > > > > +    struct rte_bbdev_op_fft {
> > > > > +        struct rte_bbdev_op_data base_input;
> > > > > +        struct rte_bbdev_op_data base_output;
> > > > > +        struct rte_bbdev_op_data power_meas_output;
> > > > > +        uint32_t op_flags;
> > > > > +        uint16_t input_sequence_size;
> > > > > +        uint16_t input_leading_padding;
> > > > > +        uint16_t output_sequence_size;
> > > > > +        uint16_t output_leading_depadding;
> > > > > +        uint8_t window_index[RTE_BBDEV_MAX_CS_2];
> > > > > +        uint16_t cs_bitmap;
> > > > > +        uint8_t num_antennas_log2;
> > > > > +        uint8_t idft_log2;
> > > > > +        uint8_t dft_log2;
> > > > > +        int8_t cs_time_adjustment;
> > > > > +        int8_t idft_shift;
> > > > > +        int8_t dft_shift;
> > > > > +        uint16_t ncs_reciprocal;
> > > > > +        uint16_t power_shift;
> > > > > +        uint16_t fp16_exp_adjust;
> > > > > +    };
> > > >
> > > > Why is this codeblock added in this guide? Isn't it covered in the
> > > > doxygen API doc?
> > >
> > > It is but here this detailed in context and with additional details.
> > > Note that this is the exact same format being used for all the other
> > > existing operations.
> > > Are you okay to keep as is?
> >
> > I would suggest to remove it and others as well. I think we can add more 
> > info
> > In the API guide if needed.
> > Otherwise it will be difficult to maintain/track in future for any changes 
> > that
> > are done in these structs.
> 
> Do we really want to do this now?
> That documentation is tightly tied with the API, so the doc would always have 
> to
> be updated when the API structure is modified.
> If you read this through most of the documentation would become barely
> readable without direct pointer to the relevant structure and related 
> parameters
> (not just for FFT but also for the existing operations).
> I see also some code snippets insertions for other parallel documentation when
> relevant.
> I can see your point about avoid duplicates but in that very case it is 
> arguably
> sensible to be readable and would personally not change it in so far as I am
> concerned.
> 
> Are you okay to keep current documentation flow for now as is, and we can
> look into improving/updating documentation as a future item based on agreed
> guidance?
> 
The maintenance of this documentation is difficult in longer run.
It will be unnecessary burden for the developer to sync both code and 
documentation.

We have seen this thing in past that people tend to forget about updating 
documentation.
And we are trying to remove unnecessary code snippets from the documentation.
We should not add new such code in documentation.
I would say remove this one from this patch. And you can submit a patch later 
to remove the
rest.

Regards,
Akhil

Reply via email to