It is time to apply for the Google Summer of Code again to get Nim accepted as a mentoring organization!
Our project idea for this year is in the Nim wiki. So far it is a copy of the ideas for 2015, which still seem relevant. We have time until February 19 to finalize the application.
In the last two years we were not accepted, this might be of interest:
As mentors we have Araq, dom96 and me so far. If you have any ideas for our GSoC application or are interested in mentoring as well, you can get in touch with me.
Is there a Nim champion inside Google who could usher this through?
Getting the LLVM backend integrated into Nim strikes me as an impactful project, though the old Nim GSOC projects looked really good to me, so I'm not a good judge.
I agree @cdunn2001 that bioinformatics could be a nice area for Nim. Lots of people there use Python and Nim would be an easier switch for many who'd need to use C++ for performance. Are you interested in starting a bio-nim project?
Well i'm not an expert in anything, but maybe a binding or lib for opengl vulkan?
I know it's not out yet (vulkan that is) but a man can dream can't he?
Porting networkx would be great but amount of work is huge. Alternative - bindings to igraph.
Here is how to wrap C++ code, which is probably a great deal of what we would need to do:
@RPG, you're right; binding igraph would be a much easier project than porting the pure-Python networkx. However, graph-tool is significantly faster than igraph. It might make more sense to port the Boost Graph Library.
Using compact intsets is an interesting idea. But we really just need to choose an API. If we go with the networkx API, we can fill it in as-needed and accept contributions.
If this is an unlikely GSoC project, we should probably take this discussion elsewhere (maybe to a github issue (or a google-group).
@cdunn2001, I'm bpr on github. Add me to the org please.
If no one wants to try this as a GSoC project, then we can start a separate thread. We're not the first to voice interest in this topic here. We can try and round up a few more.
I'd be more interested in de novo library assembly, rather than binding to C and C++ libraries. By that of course I mean a pure Nim library.
If you are looking for a simple graph library in Nim: https://github.com/superfunc/charlotte
@RPG, Any idea what happened to Charlotte?
I am porting parts of networkx to pure Nim now, with a friend. We like your IntSets idea for non-multi digraphs. Please feel free to comment: