27 juillet 2018 15:57 "Dario Casalinuovo" <b.vitruvio@xxxxxxxxx> a écrit:
You have commit access. Just revert the commit, it would waste a lot less
time.
You can explain the reasons in the commit message and we can happily move on
and
rework and resubmit the patch.
Do you believe that reverting the commit abruptly looked like a much more
rude behavior for me?
Here is my view on it:
- It is certainly not perfect. The API is not easy to use, not all that well
documented despite the
efforts from Be days, and few people actually know all parts of it.
- However, we have a mostly working implementation, and it works well enough
for now, so I prefer
to
focus my efforts in other parts of the code (intel_extreme driver, ffmpeg
decoder, webkit, etc)
and live with an imperfect API for R1.
It is not a matter of wether the API is great or not, but what to focus on
right now.
As a result, I did not do much research in the state of the current API. You
did, so it is up to
you to document the limitations, and then the possible solutions. From the
discussions I followed,
you have documented your planned solutions, but communication between us
failed at some point and
I missed the part where you listed the limitations and why your new design
fixes them.
Well, as said, I invited multiple times in multiple occasions to comment on
my work. I never
received a comment from anyone.
I never had a question on why I did something in a certain way rather than
another.
I changed myself ideas on something compared to the current implementation.
For example I planned
to introduce something like automatic buffer management, but once I begun to
implemented it, I
figured out immediatly it'd be overkill. As you may see from the code, I am
resilient in
introducing new classes and I am mostly thinking how we can reduce code
duplication and make
developing media apps more easier.
Well, I am going to work on something as long as it is near what I propose.
If people want to
follow another path, I am not going to do the work for them.