Closed a2ray closed 11 months ago
Apologies for revisiting this again ... but there is some error to do with conda
and matplotlib
in packages that have PyPlot.jl as a dependency but I can't pinpoint where. On triggering a new package release of HiQGA on GitHub using Registrator @JuliaRegistrator register
I get almost exactly the same error as shown by @Octogonapus above, specifically,
ResolvePackageNotFound:
# - conda==23.1.0
where automerge fails with the error
[ Info: Installing matplotlib via the Conda matplotlib package...
[ Info: Running `conda install -y matplotlib` in root environment
ResolvePackageNotFound:
- conda==23.1.0
The automerge error is here.
I know this is not to do with my package or PyPlot, because a manual install with
pkg>add https://github.com/GeoscienceAustralia/HiQGA.jl.git
works fine on Julia v 1.9, the same Julia version as in the GitHub/Registrator automerge script. However, new package releases with PyPlot.jl as a dependency will break, I think.
Error while trying to register: Register Failed @a2ray, it looks like you are not a publicly listed member/owner in the parent organization (JuliaPy). If you are a member/owner, you will need to change your membership to public. See GitHub Help
Error while trying to register: Register Failed @a2ray, it looks like you are not a publicly listed member/owner in the parent organization (JuliaPy). If you are a member/owner, you will need to change your membership to public. See GitHub Help
Very strange ... I fixed it on my end but had to use a build script in deps/build.jl
and do a Conda.pip
to install matplotlib
before installing through Conda.add
as normal. Otherwise, it kept failing to find conda-forge. I had never needed a build.jl
before for PyPlot
as a dependency. See here https://github.com/GeoscienceAustralia/HiQGA.jl/commit/db833bf32840503ee3bd0909b2b92993c239413c
Apologies for re-visiting the error in the closed issue https://github.com/JuliaPy/PyPlot.jl/issues/568#issuecomment-1636264150, with the error underneath
I believe that this error will impede package registration with PyPlot.jl as a dependency. Please see comments underneath.