Internal compiler error: type error in adjust_effects

[maleadt]

A couple of packages have started running into the following (non-fatal) compiler error:

TypeError(func=:if, context="", expected=Bool, got=nothing)
ijl_type_error_rt at /source/src/rtutils.c:119
ijl_type_error at /source/src/rtutils.c:127
adjust_effects at ./compiler/typeinfer.jl:457
finish at ./compiler/typeinfer.jl:568
_typeinf at ./compiler/typeinfer.jl:257
typeinf at ./compiler/typeinfer.jl:215
typeinf_edge at ./compiler/typeinfer.jl:925
abstract_call_method at ./compiler/abstractinterpretation.jl:656
abstract_call_gf_by_type at ./compiler/abstractinterpretation.jl:154
abstract_call_known at ./compiler/abstractinterpretation.jl:2057
abstract_call at ./compiler/abstractinterpretation.jl:2133
abstract_call at ./compiler/abstractinterpretation.jl:2107
abstract_eval_statement_expr at ./compiler/abstractinterpretation.jl:2279
abstract_eval_statement at ./compiler/abstractinterpretation.jl:2528
abstract_eval_basic_statement at ./compiler/abstractinterpretation.jl:2819
typeinf_local at ./compiler/abstractinterpretation.jl:2995
typeinf_nocycle at ./compiler/abstractinterpretation.jl:3083
_typeinf at ./compiler/typeinfer.jl:244
typeinf at ./compiler/typeinfer.jl:215
typeinf_ext at ./compiler/typeinfer.jl:1050
typeinf_ext_toplevel at ./compiler/typeinfer.jl:1083
typeinf_ext_toplevel at ./compiler/typeinfer.jl:1079
jfptr_typeinf_ext_toplevel_20125.clone_1 at /opt/julia/lib/julia/sys.so (unknown line)

Example: https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_date/2023-01/16/ConstraintLearning.primary.log

Bisected to #48246, cc @Keno. Putting this on the milestone as this was back-ported to 1.9, but technically it's not a fatal error so could be fixed later.

[gbaraldi]

I imagine you can reproduce locally, could you run creduce on it?

[maleadt]

It's a large package, so reducing will take a while. I think the issue should be clear to people familiar with that code (it's testing a nothing value). @aviatesk can you take a look since you reviewed that PR?

[gbaraldi]

Oh, I didn't see that line. I guess it's probably not too hard then.