Currently if a dependency is from GH via Remotes (or any way, really?), and there is a locally installed package, pak will always update that package from GH. I think this is because
- It is hard to say what the user really wants, and after all the user said they wanted that package from GH.
- It is not possible to determine the relationship between the installed package and the package in the GH repo.
Which all makes sense to me, but it also makes local development cumbersome when you are working on two packages locally and one depends on the other one.
This also happens in Positron, which uses local_install*() (?).
Btw. this is not unique to github::, but applies to all other remote types (git, gitlab, url, etc. possibly even standard, cran, bioc.)
Currently if a dependency is from GH via
Remotes(or any way, really?), and there is a locally installed package, pak will always update that package from GH. I think this is becauseWhich all makes sense to me, but it also makes local development cumbersome when you are working on two packages locally and one depends on the other one.
This also happens in Positron, which uses
local_install*()(?).Btw. this is not unique to
github::, but applies to all other remote types (git, gitlab, url, etc. possibly even standard, cran, bioc.)