Re: Help in JSON encode/decode project

[Kai Torben Ohlhus]

On 3/12/20 5:15 AM, Abdallah Elshamy wrote:
>> Based on you impressions, you can then prepare your proposal.  Of course
>> you don't need to complete anything when applying for the project.
> 
> The list seems great but I would like to add more details to clarify
> some points for me:
> 
>  1. for the present time, I should spend more time with the four
>     libraries (I already started) to decide my approach to the project
>     so I can provide a more detailed time line and prepare my proposal.
> 
>  2. During the community bonding period, I should (besides getting more
>     familiar with the organization) put a fine detailed and complete
>     plan for the functions and the test suite.
> 
>  3. The first output of my work will be a complete test suit extracted
>     from the four approaches and my additions to it. It will be
>     Matlab-compatible for benchmarking.
> 
>  4. Asses comprehensively all four libraries with your tests and
>     benchmarks. Create reliable figures, graphics to decide back-end (I
>     don’t think that this is necessary since the link that Mr.Andrew
>     Janke shared [1] provides a strong evidence that Rapidjson is better
>     is my assumption is correct?)
> 
>  5. Produce a c++ implementation for jsondecode/jsonencode.
> 
>  6. convert my test suite to Octave BIST.
> 
>  7. Provide a proper documentation for the functions.
> 
>  8. Integrate the functions into octave core.
> 
>  9. Use the community and mentors feedback after submissions to perfect
>     the patch.
> 
> A milestone list I have in mind is :
> 
>  1. deliver test suite
> 
>  2. deliver jsondecode
> 
>  3. deliver jsonencode
> 
> 
> Please, let me know what you think about this.
> 
> 
> Thanks for your time,
> 
> Abdallah
> 
> 
> [1] https://github.com/miloyip/nativejson-benchmark
> 


Looks all good to me.  You are right, I think RapidJSON should be the
choice.  Makes the project easier, but still lots of work to do ;-)
Because two of the four JSON projects now use RapidJSON, the focus of
the benchmark should not only be speed, but also Matlab compatible data
processing (like octave-jsonstuff passes 100 of 100 Matlab compatible
tests...).  This might also help those projects to improve their code.
Their focus is not only 100% Matlab compatibility, they also provide
other useful extension.  But for Octave compatibility should be more the
focus.

Best,
Kai