Python package works on Julia 1.11 and 1.13 but errors on 1.12

[nhz2]

Affects: PythonCall

Describe the bug
I expect the following output when using imagecodecs:

using Pkg
Pkg.activate(;temp=true)
pkg"add PythonCall CondaPkg"
using CondaPkg
pkg"conda pip_add imagecodecs"
using PythonCall
imcd = pyimport("imagecodecs")
imcd.lzma_encode(UInt8[])
# Python: b'\xfd7zXZ\x00\x00\x04\xe6\xd6\xb4F\x00\x00\x00\x00\x1c\xdfD!\x1f\xb6\xf3}\x01\x00\x00\x00\x00\x04YZ'

This works on Julia 1.11.8 and Julia 1.13.0-alpha2, but on Julia 1.12.3 I get:

julia> imcd.lzma_encode(UInt8[])
ERROR: Python: LzmaError: lzma_code returned LZMA_PROG_ERROR
Python stacktrace:
 [1] imagecodecs._lzma.lzma_encode
   @ imagecodecs/_lzma.pyx:137
Stacktrace:
 [1] pythrow()
   @ PythonCall.Core ~/.julia/packages/PythonCall/83z4q/src/Core/err.jl:77
 [2] errcheck
   @ ~/.julia/packages/PythonCall/83z4q/src/Core/err.jl:10 [inlined]
 [3] pycallargs(f::Py, args::Py)
   @ PythonCall.Core ~/.julia/packages/PythonCall/83z4q/src/Core/builtins.jl:194
 [4] pycall(f::Py, args::Vector{UInt8}; kwargs::@Kwargs{})
   @ PythonCall.Core ~/.julia/packages/PythonCall/83z4q/src/Core/builtins.jl:213
 [5] pycall
   @ ~/.julia/packages/PythonCall/83z4q/src/Core/builtins.jl:203 [inlined]
 [6] (::Py)(args::Vector{UInt8})
   @ PythonCall.Core ~/.julia/packages/PythonCall/83z4q/src/Core/Py.jl:363
 [7] top-level scope
   @ REPL[8]:1

Your system

julia> versioninfo()
Julia Version 1.12.3
Commit 966d0af0fdf (2025-12-15 11:20 UTC)
Build Info:
  Official https://julialang.org release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 16 × AMD Ryzen 7 9800X3D 8-Core Processor
  WORD_SIZE: 64
  LLVM: libLLVM-18.1.7 (ORCJIT, znver5)
  GC: Built with stock GC
Threads: 1 default, 0 interactive, 1 GC (on 16 virtual cores)
Environment:
  JULIA_NUM_THREADS = 1

julia> Pkg.status()
Status `/tmp/jl_bFSaIh/Project.toml`
  [992eb4ea] CondaPkg v0.2.33
  [6099a3de] PythonCall v0.9.31

julia> CondaPkg.status()
CondaPkg Status /tmp/jl_bFSaIh/CondaPkg.toml
Environment
  /tmp/jl_bFSaIh/.CondaPkg/.pixi/envs/default
Pip Packages
  imagecodecs v2025.11.11

[cjdoris]

What is pytype(Py(UInt8[])) in the working/non-working cases.

Do you still see the error if you start Julia with -t1?

Is the version of imagecodecs the same in the working/non-working cases?

[nhz2]

julia> pytype(Py(UInt8[]))
Python: <class 'juliacall.VectorValue'>

In both cases. I have JULIA_NUM_THREADS = 1 env variable set.

Interestingly, it sometimes does work on 1.12.3 if I repeat the import four times:

using Pkg
Pkg.activate(;temp=true)
pkg"add PythonCall CondaPkg"
using CondaPkg
pkg"conda pip_add imagecodecs"
using PythonCall
imcd = pyimport("imagecodecs")
imcd = pyimport("imagecodecs")
imcd = pyimport("imagecodecs")
imcd = pyimport("imagecodecs")
imcd.lzma_encode(UInt8[])

[cjdoris]

Ok that's very strange. That hints at a more fundamental bug.

Try this?

x = Py(UInt8[])
imcd.lzma_encode(x)