Stop exporting load

docs/src/mpi.md

@@ -54,7 +54,7 @@ To analyze the output, see the documentation page on [post-processing for MPI ru
 back to your interactive chain via: 
 
 ```@example local
-pt = load(result) # possible thanks to 'pigeons(..., checkpoint = true)' used above
+pt = Pigeons.load(result) # possible thanks to 'pigeons(..., checkpoint = true)' used above
 ```
 
 ## Running MPI on a cluster
@@ -115,7 +115,7 @@ To analyze the output, see the documentation page on [post-processing for MPI ru
 back to your interactive chain via: 
 
 ```
-pt = load(mpi_run) # possible thanks to 'pigeons(..., checkpoint = true)' used above
+pt = Pigeons.load(mpi_run) # possible thanks to 'pigeons(..., checkpoint = true)' used above
 ```
 
 

docs/src/output-mpi-postprocessing.md

@@ -21,7 +21,7 @@ however, when running a job via [`MPI`](@ref) or [`ChildProcess`](@ref),
 [`pigeons()`](@ref) returns a [`Result`](@ref) struct (which only holds the  
 directory where samples are stored). 
 
-Use [`load()`](@ref) to convert a [`Result`](@ref) into a 
+Use `Pigeons.load(..)` to convert a [`Result`](@ref) into a 
 [`PT`](@ref) struct. 
 This will load the information distributed across several machines 
 into the interactive node.
@@ -59,7 +59,7 @@ pt_result = pigeons(target = an_unidentifiable_model,
                 record = [traces; round_trip; record_default()])
 
 # (*) load the result across all machines into this interactive node
-pt = load(pt_result)
+pt = Pigeons.load(pt_result)
 
 # collect the statistics and convert to MCMCChains' Chains
 # to have axes labels matching variable names in Turing and Stan

src/Pigeons.jl

@@ -76,7 +76,8 @@ export pigeons, Inputs, PT,
     index_process, swap_acceptance_pr, log_sum_ratio, online, round_trip, energy_ac1, traces, disk,
     record_online, record_default,
     # utils to run on scheduler:
-    Result, load, setup_mpi, queue_status, queue_ncpus_free, kill_job, watch,
+    Result, setup_mpi, queue_status, queue_ncpus_free, kill_job, watch,
+    # load, <- removed to avoid clash - see https://github.com/Julia-Tempering/Pigeons.jl/issues/200
     # getting information out of an execution:
     stepping_stone, n_tempered_restarts, n_round_trips, process_sample, get_sample,
     # variational references:

src/pt/checks.jl

@@ -6,7 +6,7 @@ function preflight_checks(inputs::Inputs)
               """
     end
     if mpi_active() && !inputs.checkpoint
-        @warn "To be able to call load() to retrieve samples in-memory, use pigeons(target = ..., checkpoint = true)"
+        @warn "To be able to call Pigeons.load() to retrieve samples in-memory, use pigeons(target = ..., checkpoint = true)"
     end
     if Threads.nthreads() > 1 && !inputs.multithreaded
         @warn "More than one threads are available, but explore!() loop is not parallelized as inputs.multithreaded == false"

test/test_checkpoint.jl

@@ -11,7 +11,7 @@ end
         compare_pts(p1, p2)
 
         r = pigeons(;target, checkpoint = true, on = ChildProcess(n_local_mpi_processes = 2, dependencies=[DynamicPPL,]))
-        p3 = load(r)
+        p3 = Pigeons.load(r)
         compare_pts(p1, p3)
     end
 end

test/test_lazy_target.jl

@@ -10,7 +10,7 @@ include("supporting/lazy.jl")
                             n_threads = 2,
                             dependencies = ["$(@__DIR__)/supporting/lazy.jl"]
                     ))
-    pt1 = load(r)
+    pt1 = Pigeons.load(r)
     pt2 = pigeons(target = toy_mvn_target(1))
 
     @test Pigeons.recursive_equal(pt1.replicas, pt2.replicas)

test/test_resume.jl

@@ -19,14 +19,14 @@ end
     pt = pigeons(; target = toy_mvn_target(1), checkpoint = true, on = ChildProcess(n_local_mpi_processes = 2))
     exec = increment_n_rounds!(pt.exec_folder, 2)
     r = pigeons(exec, ChildProcess(n_local_mpi_processes = 2))
-    Pigeons.check_against_serial(load(r))
+    Pigeons.check_against_serial(Pigeons.load(r))
 end
 
 @testset "Extend number of rounds with PT object, on ChildProcess" begin
     pt = pigeons(; target = toy_mvn_target(1), checkpoint = true)
     pt = increment_n_rounds!(pt, 2)
     r = pigeons(pt.exec_folder, ChildProcess(n_local_mpi_processes = 2))
-    Pigeons.check_against_serial(load(r))
+    Pigeons.check_against_serial(Pigeons.load(r))
 end
 
 @testset "Complex example of increasing number of rounds many times" begin
@@ -59,5 +59,5 @@ end
     result = pigeons(new_exec_folder, ChildProcess(n_local_mpi_processes = 2))
 
     # make sure it is equivalent to doing it in one shot
-    Pigeons.check_against_serial(load(result))
+    Pigeons.check_against_serial(Pigeons.load(result))
 end

test/test_traces.jl

@@ -48,7 +48,7 @@ end
                     multithreaded = false,  # setting to true puts too much pressure on CI instances? https://github.com/Julia-Tempering/Pigeons.jl/actions/runs/5627897144/job/15251121621?pr=90
                     checkpoint = true,
                     on = ChildProcess(n_local_mpi_processes = 2, n_threads = 1, dependencies=[DynamicPPL, BridgeStan])) # setting to more than 1 puts too much pressure on CI instances?
-            pt = load(r)
+            pt = Pigeons.load(r)
             @test length(pt.reduced_recorders.traces) == 1024 * (extended_traces ? 10 : 1)
             for chain in Pigeons.chains_with_samples(pt)
                 marginal = [get_sample(pt, chain, i)[1] for i in 1:1024]