You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In pkgimage.mk, such as here.
2.5. In sysimg.jl, such as here (Note, this one counts as 1/2 because only topological sort order is recorded, rather than a full dependency graph).
This makes it needlessly difficult to alter stdlib dependencies. Instead, we should automatically compute this from the Project.toml files in all cases. We should also record the sysimg/pkgimage decision in the same place, perhaps by setting a preference in the Project.toml?
The text was updated successfully, but these errors were encountered:
#51368 did part of this and after #51399 we have much fewer listed in sysimg.jl.
For the last step we do need a toml parser and then convert that to Makefiles variables. That seems obnoxious. Or we have something that autogenerated stdlib.mk for us...
We should have a single source of truth for stdlib dependency knowledge. As it stands, we have 2.5 separate places where this knowledge is encoded:
Project.toml
such as herepkgimage.mk
, such as here.2.5. In
sysimg.jl
, such as here (Note, this one counts as1/2
because only topological sort order is recorded, rather than a full dependency graph).This makes it needlessly difficult to alter stdlib dependencies. Instead, we should automatically compute this from the Project.toml files in all cases. We should also record the sysimg/pkgimage decision in the same place, perhaps by setting a preference in the
Project.toml
?The text was updated successfully, but these errors were encountered: