Open MasonProtter opened 10 months ago
Should be a good first-issue for someone looking to make a PR (I don't have permission to add labels or I'd label it myself)
@MasonProtter I think it's the triage permission that lets one label issues. Viral has sent you an invite. Could you please check if you have received the GitHub invite on your email (or a notification)?
Yep, got it, thank you both!
Hi, is someone working on this already, else I'd love to work on it . Thanks !
I don't think so, please go for it!
Hi, It's my first time working on open source and I'm still familiarizing myself with Julia. Could you please help me out in proceeding to solve the issue. Thanks a lot
You would need to clone the Julia repo, and add the suggested code in the top comment (with appropriate modifications) to stdlib/LinearAlgebra/src/special.jl
. You would also need to add tests for each added method to stdlib/LinearAlgebra/test/special.jl
. Please create a separate @testset
mentioning the issue number (you should be able to figure this out from the file). Please ensure that the tests are running. Finally, commit your changes to a new branch (preferably not the master), fork the Julia repo and push the changes to your fork, and open a PR to the main julia repo.
Just a note for anyone else reading, we generally do not 'reserve' issues for people to fix. So if someone says they want to work on something, but doesn't actually start working on it, or if their attempt to fix it stalls out, or if you have a different, better way of solving the issues, it's completely okay to just go and submit your own PR to fix the issue.
This is not to say that people should feel encouraged to try and 'beat' someone else to submitting a PR for a feature, that's also bad behaviour. It's just that you shouldn't feel shy about working on something that someone expressed the intention to work on, but hasn't completed.
So for the various array wrapper types like
Symmetric
,Hermitian
, andAdjoint
, we currently don't have a specializedzero
method, but instead fall back to theAbstractArray
definition which works in terms ofsimilar
. That means that e.g. this gives aMMatrix
:I think though for these wrapper types, we should be able to define
zero
(and maybe some other functions?) liketo avoid falling back to
similar
where possible, and make sure that custom methods likezero(::SMatrix)
end up getting called instead.