Try #625:

libs/pika/executors/tests/unit/standalone_thread_pool_executor.cpp

@@ -158,7 +158,7 @@ int main()
 {
     {
         // Choose a scheduler.
-        using sched_type = pika::threads::local_priority_queue_scheduler<>;
+        using sched_type = pika::threads::detail::local_priority_queue_scheduler<>;
 
         // Choose all the parameters for the thread pool and scheduler.
         std::size_t const num_threads =

libs/pika/init_runtime/src/init_runtime.cpp

@@ -234,7 +234,7 @@ namespace pika {
             }
 #endif
 #ifdef PIKA_HAVE_THREAD_MINIMAL_DEADLOCK_DETECTION
-            threads::set_minimal_deadlock_detection_enabled(
+            threads::detail::set_minimal_deadlock_detection_enabled(
                 cmdline.rtcfg_.enable_minimal_deadlock_detection());
 #endif
 #ifdef PIKA_HAVE_SPINLOCK_DEADLOCK_DETECTION

libs/pika/resource_partitioner/examples/guided_pool_test.cpp

@@ -27,7 +27,7 @@ static int pool_threads = 0;
 #define CUSTOM_POOL_NAME "Custom"
 
 // this is our custom scheduler type
-using high_priority_sched = pika::threads::shared_priority_queue_scheduler<>;
+using high_priority_sched = pika::threads::detail::shared_priority_queue_scheduler<>;
 using pika::threads::scheduler_mode;
 
 // Force an instantiation of the pool type templated on our custom scheduler

libs/pika/resource_partitioner/examples/oversubscribing_resource_partitioner.cpp

@@ -43,7 +43,7 @@ static int pool_threads = 1;
 static std::string const pool_name = "mpi";
 
 // this is our custom scheduler type
-using high_priority_sched = pika::threads::shared_priority_queue_scheduler<>;
+using high_priority_sched = pika::threads::detail::shared_priority_queue_scheduler<>;
 using namespace pika::threads;
 using pika::threads::scheduler_mode;
 

libs/pika/resource_partitioner/examples/simple_resource_partitioner.cpp

@@ -33,7 +33,7 @@ static std::string const pool_name = "mpi";
 
 // ------------------------------------------------------------------------
 // this is our custom scheduler type
-using numa_scheduler = pika::threads::shared_priority_queue_scheduler<>;
+using numa_scheduler = pika::threads::detail::shared_priority_queue_scheduler<>;
 using namespace pika::threads;
 
 // ------------------------------------------------------------------------

libs/pika/schedulers/docs/index.rst

@@ -14,9 +14,9 @@ schedulers
 This module provides schedulers used by thread pools in the
 :ref:`modules_thread_pools` module. There are currently three main schedulers:
 
-* :cpp:class:`pika::threads::local_priority_queue_scheduler`
-* :cpp:class:`pika::threads::static_priority_queue_scheduler`
-* :cpp:class:`pika::threads::shared_priority_queue_scheduler`
+* :cpp:class:`pika::threads::detail::local_priority_queue_scheduler`
+* :cpp:class:`pika::threads::detail::static_priority_queue_scheduler`
+* :cpp:class:`pika::threads::detail::shared_priority_queue_scheduler`
 
 Other schedulers are specializations or variations of the above schedulers. See
 the examples of the :ref:`modules_resource_partitioner` module for examples of

libs/pika/schedulers/include/pika/schedulers/deadlock_detection.hpp

@@ -9,9 +9,9 @@
 
 #include <pika/config.hpp>
 
-namespace pika::threads {
+namespace pika::threads::detail {
 #ifdef PIKA_HAVE_THREAD_MINIMAL_DEADLOCK_DETECTION
     PIKA_EXPORT void set_minimal_deadlock_detection_enabled(bool enabled);
     PIKA_EXPORT bool get_minimal_deadlock_detection_enabled();
 #endif
-}    // namespace pika::threads
+}    // namespace pika::threads::detail

libs/pika/schedulers/include/pika/schedulers/local_priority_queue_scheduler.hpp

@@ -41,7 +41,7 @@
 // TODO: add branch prediction and function heat
 
 ///////////////////////////////////////////////////////////////////////////////
-namespace pika::threads {
+namespace pika::threads::detail {
     ///////////////////////////////////////////////////////////////////////////
 #if defined(PIKA_HAVE_CXX11_STD_ATOMIC_128BIT)
     using default_local_priority_queue_scheduler_terminated_queue = lockfree_lifo;
@@ -60,7 +60,7 @@ namespace pika::threads {
     template <typename Mutex = std::mutex, typename PendingQueuing = lockfree_fifo,
         typename StagedQueuing = lockfree_fifo,
         typename TerminatedQueuing = default_local_priority_queue_scheduler_terminated_queue>
-    class PIKA_EXPORT local_priority_queue_scheduler : public detail::scheduler_base
+    class PIKA_EXPORT local_priority_queue_scheduler : public scheduler_base
     {
     public:
         using has_periodic_maintenance = std::false_type;
@@ -76,7 +76,7 @@ namespace pika::threads {
         {
             init_parameter(std::size_t num_queues, pika::detail::affinity_data const& affinity_data,
                 std::size_t num_high_priority_queues = std::size_t(-1),
-                detail::thread_queue_init_parameters thread_queue_init = {},
+                thread_queue_init_parameters thread_queue_init = {},
                 char const* description = "local_priority_queue_scheduler")
               : num_queues_(num_queues)
               , num_high_priority_queues_(num_high_priority_queues == std::size_t(-1) ?
@@ -100,15 +100,15 @@ namespace pika::threads {
 
             std::size_t num_queues_;
             std::size_t num_high_priority_queues_;
-            detail::thread_queue_init_parameters thread_queue_init_;
+            thread_queue_init_parameters thread_queue_init_;
             pika::detail::affinity_data const& affinity_data_;
             char const* description_;
         };
         using init_parameter_type = init_parameter;
 
         local_priority_queue_scheduler(
             init_parameter_type const& init, bool deferred_initialization = true)
-          : detail::scheduler_base(init.num_queues_, init.description_, init.thread_queue_init_)
+          : scheduler_base(init.num_queues_, init.description_, init.thread_queue_init_)
           , curr_queue_(0)
           , affinity_data_(init.affinity_data_)
           , num_queues_(init.num_queues_)
@@ -1262,11 +1262,11 @@ namespace pika::threads {
         std::vector<pika::concurrency::detail::cache_line_data<std::vector<std::size_t>>>
             victim_threads_;
     };
-}    // namespace pika::threads
+}    // namespace pika::threads::detail
 
 template <typename Mutex, typename PendingQueuing, typename StagedQueuing,
     typename TerminatedQueuing>
-struct fmt::formatter<pika::threads::local_priority_queue_scheduler<Mutex, PendingQueuing,
+struct fmt::formatter<pika::threads::detail::local_priority_queue_scheduler<Mutex, PendingQueuing,
     StagedQueuing, TerminatedQueuing>> : fmt::formatter<pika::threads::detail::scheduler_base>
 {
     template <typename FormatContext>

libs/pika/schedulers/include/pika/schedulers/local_queue_scheduler.hpp

@@ -39,7 +39,7 @@
 // TODO: add branch prediction and function heat
 
 ///////////////////////////////////////////////////////////////////////////////
-namespace pika::threads {
+namespace pika::threads::detail {
     ///////////////////////////////////////////////////////////////////////////
 #if defined(PIKA_HAVE_CXX11_STD_ATOMIC_128BIT)
     using default_local_queue_scheduler_terminated_queue = lockfree_lifo;
@@ -54,7 +54,7 @@ namespace pika::threads {
     template <typename Mutex = std::mutex, typename PendingQueuing = lockfree_fifo,
         typename StagedQueuing = lockfree_fifo,
         typename TerminatedQueuing = default_local_queue_scheduler_terminated_queue>
-    class PIKA_EXPORT local_queue_scheduler : public detail::scheduler_base
+    class PIKA_EXPORT local_queue_scheduler : public scheduler_base
     {
     public:
         using has_periodic_maintenance = std::false_type;
@@ -65,7 +65,7 @@ namespace pika::threads {
         struct init_parameter
         {
             init_parameter(std::size_t num_queues, pika::detail::affinity_data const& affinity_data,
-                detail::thread_queue_init_parameters thread_queue_init = {},
+                thread_queue_init_parameters thread_queue_init = {},
                 char const* description = "local_queue_scheduler")
               : num_queues_(num_queues)
               , thread_queue_init_(thread_queue_init)
@@ -84,14 +84,14 @@ namespace pika::threads {
             }
 
             std::size_t num_queues_;
-            detail::thread_queue_init_parameters thread_queue_init_;
+            thread_queue_init_parameters thread_queue_init_;
             pika::detail::affinity_data const& affinity_data_;
             char const* description_;
         };
         using init_parameter_type = init_parameter;
 
         local_queue_scheduler(init_parameter_type const& init, bool deferred_initialization = true)
-          : detail::scheduler_base(init.num_queues_, init.description_, init.thread_queue_init_)
+          : scheduler_base(init.num_queues_, init.description_, init.thread_queue_init_)
           , queues_(init.num_queues_)
           , curr_queue_(0)
           , affinity_data_(init.affinity_data_)
@@ -876,13 +876,12 @@ namespace pika::threads {
         std::vector<::pika::threads::detail::mask_type> numa_domain_masks_;
         std::vector<::pika::threads::detail::mask_type> outside_numa_domain_masks_;
     };
-}    // namespace pika::threads
+}    // namespace pika::threads::detail
 
 template <typename Mutex, typename PendingQueuing, typename StagedQueuing,
     typename TerminatedQueuing>
-struct fmt::formatter<
-    pika::threads::local_queue_scheduler<Mutex, PendingQueuing, StagedQueuing, TerminatedQueuing>>
-  : fmt::formatter<pika::threads::detail::scheduler_base>
+struct fmt::formatter<pika::threads::detail::local_queue_scheduler<Mutex, PendingQueuing,
+    StagedQueuing, TerminatedQueuing>> : fmt::formatter<pika::threads::detail::scheduler_base>
 {
     template <typename FormatContext>
     auto format(pika::threads::detail::scheduler_base const& scheduler, FormatContext& ctx)

libs/pika/schedulers/include/pika/schedulers/lockfree_queue_backends.hpp

@@ -26,7 +26,7 @@
 #include <cstdint>
 #include <utility>
 
-namespace pika::threads {
+namespace pika::threads::detail {
 
     struct lockfree_fifo;
 
@@ -331,4 +331,4 @@ namespace pika::threads {
 
 #endif    // PIKA_HAVE_CXX11_STD_ATOMIC_128BIT
 
-}    // namespace pika::threads
+}    // namespace pika::threads::detail

libs/pika/schedulers/include/pika/schedulers/maintain_queue_wait_times.hpp

@@ -9,9 +9,9 @@
 
 #include <pika/config.hpp>
 
-namespace pika::threads {
+namespace pika::threads::detail {
 #ifdef PIKA_HAVE_THREAD_QUEUE_WAITTIME
     PIKA_EXPORT void set_maintain_queue_wait_times_enabled(bool enabled);
     PIKA_EXPORT bool get_maintain_queue_wait_times_enabled();
 #endif
-}    // namespace pika::threads
+}    // namespace pika::threads::detail

libs/pika/schedulers/include/pika/schedulers/queue_holder_numa.hpp

@@ -42,12 +42,12 @@
 # endif
 #endif
 
-namespace pika {
+namespace pika::detail {
     static pika::debug::detail::enable_print<QUEUE_HOLDER_NUMA_DEBUG> nq_deb("QH_NUMA");
 }
 
 // ------------------------------------------------------------////////
-namespace pika::threads {
+namespace pika::threads::detail {
     // ----------------------------------------------------------------
     // Helper class to hold a set of thread queue holders.
     // ----------------------------------------------------------------
@@ -105,7 +105,7 @@ namespace pika::threads {
                 if (queues_[q]->get_next_thread_HP(thrd, (stealing || (i > 0)), i == 0))
                 {
                     // clang-format off
-                    nq_deb.debug(debug::detail::str<>("HP/BP get_next")
+                    pika::detail::nq_deb.debug(debug::detail::str<>("HP/BP get_next")
                          , "D", debug::detail::dec<2>(domain_)
                          , "Q",  debug::detail::dec<3>(q)
                          , "Qidx",  debug::detail::dec<3>(qidx)
@@ -135,7 +135,7 @@ namespace pika::threads {
                 // if we got a thread, return it, only allow stealing if i>0
                 if (queues_[q]->get_next_thread(thrd, (stealing || (i > 0))))
                 {
-                    nq_deb.debug(debug::detail::str<>("get_next"), "D",
+                    pika::detail::nq_deb.debug(debug::detail::str<>("get_next"), "D",
                         debug::detail::dec<2>(domain_), "Q", debug::detail::dec<3>(q), "Qidx",
                         debug::detail::dec<3>(qidx),
                         ((i == 0 && !stealing) ? "taken" : "stolen from"),
@@ -162,7 +162,7 @@ namespace pika::threads {
                 if (added > 0)
                 {
                     // clang-format off
-                    nq_deb.debug(debug::detail::str<>("HP/BP add_new")
+                    pika::detail::nq_deb.debug(debug::detail::str<>("HP/BP add_new")
                         , "added", debug::detail::dec<>(added)
                         , "D", debug::detail::dec<2>(domain_)
                         , "Q",  debug::detail::dec<3>(q)
@@ -191,7 +191,7 @@ namespace pika::threads {
                 if (added > 0)
                 {
                     // clang-format off
-                    nq_deb.debug(debug::detail::str<>("add_new")
+                    pika::detail::nq_deb.debug(debug::detail::str<>("add_new")
                          , "added", debug::detail::dec<>(added)
                          , "D", debug::detail::dec<2>(domain_)
                          , "Q",  debug::detail::dec<3>(q)
@@ -294,4 +294,4 @@ namespace pika::threads {
         void on_stop_thread(std::size_t /* num_thread */) {}
         void on_error(std::size_t /* num_thread */, std::exception_ptr const& /* e */) {}
     };
-}    // namespace pika::threads
+}    // namespace pika::threads::detail