-
-
Notifications
You must be signed in to change notification settings - Fork 359
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Repository reorganization #4359
Comments
Well, the Makie situation is a bit different, since changes in Makie.jl frequently involves changes in all backends (GLMakie, WGLMakie and CairoMakie) as well. Apart from API changes and moving of functionality we rarely need to syncronize PRs between the different parts, that were split from Plots.jl. So let me comment on the different packages in detail
I am all in of centralization of documentation in |
Also, the Makie.jl example is also a bit different since dependencies are reversed:
IMO, the current distributed structure is also preventing API changes and code evolution (which is fundamentally necessary if we wish to keep a project alive).
Let me be clearer, it will still be distinct from
I think we have tools such as tags to categorize issues and PRs. Anyway, just my two cents on the evolution of |
I know that, but you signal a kind of ownership of that package. People can only find that package going to the Plots repository, they can only gain direct acces by gaining direct access to Plots, etc... |
I think @BeastyBlacksmith 's view seems like a reasonable subset of what you propose. I've often had to work between Plots/RecipesPipeline/RecipesBase to resolve an issue, but I've not yet had to touch PlotUtils. |
Let's try with |
We should figure out how to configure TagBot to create github releases for the subdirectories and have it create the corresponding summaries based on certain labels. Currently all PRs are listed in the last Plots release. |
Not supported atm: JuliaRegistries/TagBot#157. |
Wasn't one of the reasons to have a seperate repository for RecipesBase.jl was to have a lightweight package that didn't take forever to load? While we can use the archived repository for a while but eventually code drift will become an issue. Is there a way to just load the RecipesBase.jl that lives in Plots.jl |
RecipesBase is still separated from Plots, and registered as a standalone package, so could you clarify your issue ? |
https://github.com/JuliaPlots/RecipesBase.jl is archived and is read only. While we can use it for now eventually RecipesBase.jl in Plots.jl will be updated and the archived RecipesBase.jl will break. |
Repository =/= package The RecipesBase package lives in the Plots repository, but it's still its own thing, and the registry points to it. When you do |
Plots
is kind of hard to contribute to (I take my example: I started to contribute about a year ago transitioning fromFortran
/Python
combination toJulia
), because the code is split among different packages:RecipesBase
,RecipesPipeline
,PlotUtils
,Plots
, leading to the following non-exhaustive list of problems:grep
and friends) for understanding the algorithms;Plots
related issues and PRs are split (this is not easy for maintenance, and we lack maintainers);Pkg.develop
each modified pkg, and mixing dependent PRs is sometimes cumbersome;I would thus propose to integrate
RecipesBase.jl
, ... toPlots.jl
(centralization of dependencies) as what is done inMakie
, whereGLMakie
lives inMakie.jl/GLMakie
.The rearranged repository would look like:
Plots.jl/RecipesBase
Plots.jl/RecipesPipeline
Plots.jl/PlotUtils
Plots.jl/Project.toml
Plots.jl/...
CI would be easily rearranged to test
RecipesBase
, ... first, and thenPlots
.The sub-packages would remain split (same
uuid
, we would have to change the url in theGeneral
registry), so this should be transparent to e.g.Makie
which usesPlotUtils
.The docs would have to be re-arranged too, but this shouldn't be much of a problem.
Old repositories would have to be archived (github label) and the link to the old docs removed (not sure how to do that).
I'm opening this issue to see if there are arguments that I didn't think of, which could prevent reorganization.
Cc relevant contributors: @tbreloff @BeastyBlacksmith @daschw @mkborregaard @jheinen.
The text was updated successfully, but these errors were encountered: