Subject: Importance order of Boost.GIL 3 ideas
From: Olzhas Zhumabek (anonymous.from.applecity_at_[hidden])
Date: 2019-03-26 10:34:51
I have found this page:
https://github.com/boostorg/gil/wiki/Boost.GIL-3-Ideas and would like to
know which ones have greater priority than others, in order to try it out
some and see if I should submit a proposal on that topic.
Let me list the ideas in order of my strengths:
1. Modern C++ usage
I have good grasp of C++'s mechanisms and language constructs. I often do
code reviews on usage of modern C++ and standard library, and also track
new idioms created in the community. I'm also good in template
metaprogramming. I'm not sure though how much important this is, I guess it
could be done any time, just complexity gets higher and hazards of breaking
backward compatibility rises.
2. Integration with various libraries
I have experience working with Windows APIs and tooling. Though non-Windows
libraries are not a problem too (I have Arch installed alongside Windows).
Aside from a giant named openCV, I haven't found any library that would
show performance as one of their benefits. There is also Magick++, but I
believe the main reason it is used is because ImageMagick is already
installed or easily available. I guess as long as performance is not
miserable it is ok. I'm enrolled into GPGPU computing course, so I have
some CUDA experience. I believe going in this direction might be a waste of
effort before executors land into the standard, or at least will get a bit
more solid shape. There is SYCL standard and open-source implementation of
it: https://github.com/triSYCL/triSYCL . It doesn't look too stable and it
seems like it doesn't support Nvidia GPUs, as it needs OpenCL SPIR, which
in turn requires OpenCL 1.2, which Nvidia doesn't support. Also I wouldn't
expect it to be easy to install.
I'm not sure if I'll be able to implement extra hard algorithms, but as
long as there is a specification/definition of an algorithm I will be able
to correctly implement it.
X. Proposal by Miral Shirah
I guess this one is a proposal by another participant? If not, I could
implement some more algorithms from that list and add to my submission.
Which one of those have higher priority and do they have some important
specific problems yet to be solved?
Boost list run by Boost-Gil-Owners