Current master always re-precompiles REPLExt when entering pkg>-mode in a clean environment
#53264
Comments
|
FWIW, I also got this earlier while building julia locally: Unsure whether that message is related. |
|
Things that didn't help:
|
|
I wonder if it could have to do with how your manifest looked like when you built Julia. Due to #51479 maybe it cached some bad dependency info at build time. |
|
No, that doesn't seem to be it - neither when nuking |
|
For the last week at least I have to erase |
|
┌ Debug: Rejecting cache file /home/mkitti/.julia/compiled/v1.11/REPLExt/m7ggP_iP0Qa.ji because uuid mapping for Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("d6f4376e-aef5-505a-96c1-9c027394607a"), "Markdown") has changed, The expected is wrong here. It is due to Base.identify_package(
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt"),
"Markdown"
)Lines 3459 to 3464 in a3e0b62 julia> out = open("/home/mkitti/.julia/compiled/v1.11/REPLExt/m7ggP_iP0Qa.ji", "r") do io
checksum = Base.isvalid_cache_header(io)
Base.parse_cache_header(io, "/home/mkitti/.julia/compiled/v1.11/REPLExt/m7ggP_iP0Qa.ji")
end;
julia> requires = out[2][3]
5-element Vector{Pair{Base.PkgId, Base.PkgId}}:
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("d6f4376e-aef5-505a-96c1-9c027394607a"), "Markdown")
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("cf7118a7-6976-5b1a-9a39-7adc72f591a4"), "UUIDs")
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("ade2ca70-3891-5945-98fb-dc099432e06a"), "Dates")
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("3fa0cd96-eef1-5676-8a61-b3b8758bbffb"), "REPL")
Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt") => Base.PkgId(Base.UUID("44cfe95a-1eb2-52ea-b672-e2afdf69b78f"), "Pkg")
julia> Base.identify_package(requires[1][1], requires[1][2].name)
julia> Base.identify_package(requires[1][1], requires[1][2].name) |> isnothing
true
julia> Base.identify_package(requires[2][1], requires[2][2].name) |> isnothing
true
julia> Base.identify_package(requires[3][1], requires[3][2].name) |> isnothing
true
julia> Base.identify_package(requires[4][1], requires[4][2].name) |> isnothing
true
julia> Base.identify_package(requires[5][1], requires[5][2].name) |> isnothing
true |
|
My current thinking is that Lines 316 to 318 in 4be67e4 function identify_package_env(where::PkgId, name::String)
cache = LOADING_CACHE[]
if haskey(EXT_PRIMED, where)
where = EXT_PRIMED[where]
end
if cache !== nothing
# ... |
|
The issue is that julia> REPLExt = Base.PkgId(Base.UUID("e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b"), "REPLExt")
REPLExt [e5eb5ef1-03cf-53a7-ae1d-5a66b08e832b]
julia> Base.implicit_manifest_deps_get(Base.load_path()[2], REPLExt, "Markdown")
julia> Base.implicit_manifest_deps_get(Base.load_path()[2], REPLExt, "REPL")
|
Minimum DemoUser WorkaroundThis creates an entry for Potential Solutions
|
|
If there is some corner case with implicit environments and extensions it shouldn't require much to fix that. I thought #52428 would already have handled this. I mean, Statistics have had an extension on SparseArrays.jl for quite a while so it isn't like this is something new. |
I don't think we ship a precompiled copy of that cache-file though in contrast to the REPLExt. |
We appear to also have the same issue with |
…3314) Alternative to #53293 This does kind of the same thing but it does not rely on `EXT_PRIMED` to identify extensions which I think is a better idea since we don't really want code lookup to depend on the state of Julia itself imo. Fixes #53264 ----------------------- Co-authored by: Mark Kittisopikul <`markkitti@fosstodon.org`>
Steps:
95233619745ef84c96af82299b162aa008a7debd)rmyour default environment completelyjulia -q)pkg>modeLog from my machine with `JULIA_DEBUG=loading`
The text was updated successfully, but these errors were encountered: