Remove ForwardDiff trigger

.github/workflows/ReusableTest.yml

@@ -45,6 +45,8 @@ jobs:
       - uses: julia-actions/julia-runtest@v1
       - uses: julia-actions/julia-processcoverage@v1
         if: ${{ inputs.run_codecov }}
+        with:
+          directories: src,ext
       - uses: codecov/codecov-action@v7
         if: ${{ inputs.run_codecov }}
         with:

.gitignore

@@ -1,6 +1,10 @@
 *.jl.*.cov
 *.jl.cov
 *.jl.mem
+lcov.info
+# codecov uploader binaries fetched by dev/gpu-coverage.sh
+/codecov
+/codecov-legacy
 *.rej
 .DS_Store
 .benchmarkci

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Removed
+
+- `DynamicPPLExt` no longer requires `ForwardDiff` as a triggering library to load.
+
 ## [0.2.0] - 2026-06-29
 
 ### Changed

Project.toml

@@ -23,7 +23,7 @@ Mooncake = "da2b9cff-9c12-43a0-ae48-6db2b0edb7d6"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [extensions]
-DynamicPPLExt = ["DynamicPPL", "FlexiChains", "ForwardDiff", "LogDensityProblems"]
+DynamicPPLExt = ["DynamicPPL", "FlexiChains", "LogDensityProblems"]
 EnzymeExt = "Enzyme"
 LogDensityProblemsExt = "LogDensityProblems"
 

codecov.yml

@@ -8,3 +8,11 @@ coverage:
       default:
         target: 90%
   range: "80...90" # Yellow within this range
+
+# ext/EnzymeExt.jl holds custom EnzymeRules (forward/augmented_primal/reverse).
+# Enzyme runs those rule bodies through its own compiled derivative pipeline,
+# which does not emit Julia's --code-coverage line counters, so they report as
+# uncovered even though test-Owned-Matmul.jl exercises them and asserts correct
+# numerics.
+ignore:
+  - "ext/EnzymeExt.jl"

ext/DynamicPPLExt.jl

@@ -8,7 +8,7 @@ using FlexiChains: FlexiChain, VarName, VNChain, SymChain
 using LogDensityProblems: LogDensityProblems
 
 """
-    DensityModel(turing_model::DynamicPPL.Model; ad_backend=ADTypes.AutoForwardDiff(), hvp=nothing)
+    DensityModel(turing_model::DynamicPPL.Model; ad_backend, hvp=nothing)
 
 Convenience constructor: wraps a DynamicPPL/Turing `@model` directly as a
 `DensityModel`, automatically extracting parameter names and wiring up gradient
@@ -19,22 +19,20 @@ triggers for this extension), plus any AD backend that is used.
 
 # Example
 ```julia
-using Turing, ParallelMCMC, FlexiChains
+using Turing, ParallelMCMC, FlexiChains, ForwardDiff
 
 @model function mymodel(y)
     μ ~ Normal(0, 1)
     y ~ Normal(μ, 0.5)
 end
 
-# AutoForwardDiff is the default. For larger models pass an explicit backend
-# and `using` the corresponding package (Enzyme, Mooncake).
-model = DensityModel(mymodel(1.5))
+model = DensityModel(mymodel(1.5); ad_backend=AutoForwardDiff())
 chain = sample(model, AdaptiveMALASampler(0.3; n_warmup=500), 2_000;
                chain_type=FlexiChains.VNChain, discard_warmup=true, progress=true)
 ```
 """
 function ParallelMCMC.DensityModel(
-    turing_model::DynamicPPL.Model; ad_backend=ADTypes.AutoForwardDiff(), hvp=nothing
+    turing_model::DynamicPPL.Model; ad_backend, hvp=nothing
 )
     # Sample in linked/unconstrained space and let DynamicPPL provide the gradient.
     ld = DynamicPPL.LogDensityFunction(

test/test-DEER-Turing-Logistic.jl

@@ -68,7 +68,9 @@ end
 
 function _deer_logistic_turing_density_model()
     return DensityModel(
-        _deer_logistic_regression(_LR_X, _LR_y); hvp=(β, v) -> _hvp_lr(β, v, _LR_X, _LR_y)
+        _deer_logistic_regression(_LR_X, _LR_y);
+        ad_backend=ADTypes.AutoForwardDiff(),
+        hvp=(β, v) -> _hvp_lr(β, v, _LR_X, _LR_y),
     )
 end
 

test/test-Turing-Integration.jl

@@ -70,7 +70,7 @@ end
 end
 
 @testset "DynamicPPLExt: convenience constructor" begin
-    model = DensityModel(normal_model(TRUE_OBS))
+    model = DensityModel(normal_model(TRUE_OBS); ad_backend=ADTypes.AutoForwardDiff())
 
     @test model.dim == 1
     @test isfinite(model.logdensity([0.0]))
@@ -97,15 +97,15 @@ end
 end
 
 @testset "DynamicPPLExt: convenience constructor uses linked space for constrained models" begin
-    model = DensityModel(beta_model())
+    model = DensityModel(beta_model(); ad_backend=ADTypes.AutoForwardDiff())
 
     @test model.dim == 1
     @test isfinite(model.logdensity([-0.4]))
     @test isfinite(model.grad_logdensity([-0.4])[1])
 end
 
 @testset "DynamicPPLExt: generic Turing model works with ParallelMALA and default Enzyme HVP" begin
-    model = DensityModel(normal_model(TRUE_OBS))
+    model = DensityModel(normal_model(TRUE_OBS); ad_backend=ADTypes.AutoForwardDiff())
 
     @test model.hvp === nothing
 
@@ -133,7 +133,7 @@ end
 end
 
 @testset "DynamicPPLExt: MvNormal(zeros(2), I) runs with ParallelMALA" begin
-    model = DensityModel(mvnormal_2d_model())
+    model = DensityModel(mvnormal_2d_model(); ad_backend=ADTypes.AutoForwardDiff())
 
     @test model.dim == 2
     @test isfinite(model.logdensity(zeros(2)))
@@ -163,7 +163,7 @@ end
     Dirichlet(ones(3)) lives on a 2-simplex, so its unconstrained
     representation has dim 2. Bijectors handles the link/unlink.
     =#
-    model = DensityModel(dirichlet_3_model())
+    model = DensityModel(dirichlet_3_model(); ad_backend=ADTypes.AutoForwardDiff())
 
     @test model.dim == 2
     @test isfinite(model.logdensity(zeros(2)))
@@ -184,24 +184,24 @@ end
     @test chain isa VNChain
     @test all(isfinite, Array(chain))
 end
-
-@testset "DynamicPPLExt: ParallelMALA bundle_samples fallback path (thinning)" begin
-    #= A non-default kwarg (here `thinning`) forces ParallelMALA's `mcmcsample` override =#
-    model = DensityModel(mvnormal_2d_model())
-    sampler = ParallelMALASampler(
-        0.2; T=8, maxiter=80, tol_abs=1e-4, tol_rel=1e-3, backend=ADTypes.AutoEnzyme()
-    )
-    chain = sample(
-        MersenneTwister(3), model, sampler, 800;
-        initial_params=zeros(2), chain_type=VNChain, thinning=2, progress=false,
-    )
-    @test chain isa VNChain
-    @test only(FlexiChains.parameters(chain)) == @varname(x)
-    @test all(isfinite, Array(chain))
-end
-
+
+@testset "DynamicPPLExt: ParallelMALA bundle_samples fallback path (thinning)" begin
+    #= A non-default kwarg (here `thinning`) forces ParallelMALA's `mcmcsample` override =#
+    model = DensityModel(mvnormal_2d_model(); ad_backend=ADTypes.AutoForwardDiff())
+    sampler = ParallelMALASampler(
+        0.2; T=8, maxiter=80, tol_abs=1e-4, tol_rel=1e-3, backend=ADTypes.AutoEnzyme()
+    )
+    chain = sample(
+        MersenneTwister(3), model, sampler, 800;
+        initial_params=zeros(2), chain_type=VNChain, thinning=2, progress=false,
+    )
+    @test chain isa VNChain
+    @test only(FlexiChains.parameters(chain)) == @varname(x)
+    @test all(isfinite, Array(chain))
+end
+
 @testset "DynamicPPLExt: named columns in Chains output" begin
-    model = DensityModel(normal_model(TRUE_OBS))
+    model = DensityModel(normal_model(TRUE_OBS); ad_backend=ADTypes.AutoForwardDiff())
 
     chain = sample(
         MersenneTwister(2),
@@ -219,7 +219,7 @@ end
 end
 
 @testset "discard_warmup=true removes warmup samples" begin
-    model = DensityModel(normal_model(TRUE_OBS))
+    model = DensityModel(normal_model(TRUE_OBS); ad_backend=ADTypes.AutoForwardDiff())
     n_warmup = 200
     n_total = 800
     sampler = AdaptiveMALASampler(0.3; n_warmup=n_warmup)
@@ -248,7 +248,7 @@ end
 end
 
 @testset "posterior mean and variance match analytic solution" begin
-    model = DensityModel(normal_model(TRUE_OBS))
+    model = DensityModel(normal_model(TRUE_OBS); ad_backend=ADTypes.AutoForwardDiff())
     n_warmup = 2_000
     n_draw = 10_000
     sampler = AdaptiveMALASampler(0.3; n_warmup=n_warmup)
@@ -271,7 +271,7 @@ end
 
 @testset "multivariate model: named columns for each dimension" begin
     obs = [1.0, -1.0]
-    model = DensityModel(mv_model(obs))
+    model = DensityModel(mv_model(obs); ad_backend=ADTypes.AutoForwardDiff())
 
     # 2 from linked Dirichlet + 2 from product_distribution
     @test model.dim == 4