Formatting updates.

Author: alliepiper

.clang-format

@@ -41,16 +41,28 @@ CompactNamespaces: false
 ContinuationIndentWidth: 2
 IncludeBlocks:   Regroup
 IncludeCategories:
-  - Regex:           '^<cub'
+  - Regex:           '^<nvbench'
     Priority:        1
-  - Regex:           '^<thrust'
+  - Regex:           '^<cub'
     Priority:        2
-  - Regex:           '^<cuda'
+  - Regex:           '^<thrust'
     Priority:        3
-  - Regex:           '^<nvbench'
+  - Regex:           '^<cuda/'
     Priority:        4
-  - Regex:           '^<[a-z]*>$'
+  - Regex:           '^<cuda'
     Priority:        5
+  - Regex:           '^<nvml'
+    Priority:        6
+  - Regex:           '^<cupti'
+    Priority:        7
+  - Regex:           '^<nvperf'
+    Priority:        8
+  - Regex:           '^<nlohmann'
+    Priority:        9
+  - Regex:           '^<fmt'
+    Priority:        10
+  - Regex:           '^<[a-z_]*>$'
+    Priority:        11
 IndentCaseLabels: true
 IndentPPDirectives: None
 IndentWidth: 2

.devcontainer/README.md

@@ -34,7 +34,7 @@ CCCL uses [Development Containers](https://containers.dev/) to provide consisten
 
    - Alternatively, use the Command Palette to start a Dev Container. Press `Ctrl+Shift+P` to open the Command Palette. Type "Remote-Containers: Reopen in Container" and select it.
 
-     ![Shows "Reopen in Container" in command pallete.](./img/open_in_container_manual.png)
+     ![Shows "Reopen in Container" in command palette.](./img/open_in_container_manual.png)
 
 4. Select an environment with the desired CTK and host compiler from the list:
 
@@ -136,7 +136,7 @@ For more information, see the `.devcontainer/make_devcontainers.sh --help` messa
 
 2. Install WSL 2 by running:
 ```bash
-wsl --install 
+wsl --install
 ```
 This should probably install Ubuntu distro as a default.
 
@@ -182,14 +182,14 @@ then run `sudo systemctl restart docker.service`.
 10. Open the CCCL cloned repo in VS Code ( `Ctrl + Shift + P `, select `File: Open Folder...` and select the path where your CCCL clone is located).
 
 11. If prompted, choose `Reopen in Container`.
-    
+
     - If you are not prompted just type `Ctrl + Shift + P` and `Dev Containers: Open Folder in Container ...`.
 
 12. Verify that Dev Container was configured properly by running `nvidia-smi` in your Dev Container terminal. For a proper configuration it is important for the steps in [Install prerequisites and VS Code extensions](#prereqs) to be followed in a precise order.
 
 From that point on, the guide aligns with our [existing Dev Containers native Linux guide](https://github.com/NVIDIA/cccl/blob/main/.devcontainer/README.md) with just one minor potential alteration:
 
-13. If WSL was launched without the X-server enabled, when asked to "authenticate Git with your Github credentials", if you answer **Yes**, the browser might not open automatically, with the following error message. 
+13. If WSL was launched without the X-server enabled, when asked to "authenticate Git with your Github credentials", if you answer **Yes**, the browser might not open automatically, with the following error message.
 
 > Failed opening a web browser at https://github.com/login/device
   exec: "xdg-open,x-www-browser,www-browser,wslview": executable file not found in $PATH

.devcontainer/launch.sh

@@ -304,4 +304,3 @@ main() {
 }
 
 main "$@"
-

.git-blame-ignore-revs

@@ -14,4 +14,3 @@
 #
 # Only add commits that are pure formatting changes (e.g. clang-format version changes, etc).
 8f1152d4a22287a35be2dde596e3cf86ace8054a # Increase column limit to 100
-

ci/ninja_summary.py

@@ -65,6 +65,7 @@
 
 class Target:
     """Represents a single line read for a .ninja_log file."""
+
     def __init__(self, start, end):
         """Creates a target object by passing in the start/end times in seconds
         as a float."""
@@ -94,20 +95,20 @@ def WeightedDuration(self):
         """
         # Allow for modest floating-point errors
         epsilon = 0.000002
-        if (self.weighted_duration > self.Duration() + epsilon):
-            print('%s > %s?' % (self.weighted_duration, self.Duration()))
-        assert (self.weighted_duration <= self.Duration() + epsilon)
+        if self.weighted_duration > self.Duration() + epsilon:
+            print("%s > %s?" % (self.weighted_duration, self.Duration()))
+        assert self.weighted_duration <= self.Duration() + epsilon
         return self.weighted_duration
 
     def DescribeTargets(self):
         """Returns a printable string that summarizes the targets."""
         # Some build steps generate dozens of outputs - handle them sanely.
         # The max_length was chosen so that it can fit most of the long
         # single-target names, while minimizing word wrapping.
-        result = ', '.join(self.targets)
+        result = ", ".join(self.targets)
         max_length = 65
         if len(result) > max_length:
-            result = result[:max_length] + '...'
+            result = result[:max_length] + "..."
         return result
 
 
@@ -121,12 +122,11 @@ def ReadTargets(log, show_all):
     # targets.
     if not header:
         return []
-    assert header == '# ninja log v5\n', \
-           'unrecognized ninja log version %r' % header
+    assert header == "# ninja log v5\n", "unrecognized ninja log version %r" % header
     targets_dict = {}
     last_end_seen = 0.0
     for line in log:
-        parts = line.strip().split('\t')
+        parts = line.strip().split("\t")
         if len(parts) != 5:
             # If ninja.exe is rudely halted then the .ninja_log file may be
             # corrupt. Silently continue.
@@ -165,17 +165,17 @@ def ReadTargets(log, show_all):
 def GetExtension(target, extra_patterns):
     """Return the file extension that best represents a target.
 
-  For targets that generate multiple outputs it is important to return a
-  consistent 'canonical' extension. Ultimately the goal is to group build steps
-  by type."""
+    For targets that generate multiple outputs it is important to return a
+    consistent 'canonical' extension. Ultimately the goal is to group build steps
+    by type."""
     for output in target.targets:
         if extra_patterns:
-            for fn_pattern in extra_patterns.split(';'):
-                if fnmatch.fnmatch(output, '*' + fn_pattern + '*'):
+            for fn_pattern in extra_patterns.split(";"):
+                if fnmatch.fnmatch(output, "*" + fn_pattern + "*"):
                     return fn_pattern
         # Not a true extension, but a good grouping.
-        if output.endswith('type_mappings'):
-            extension = 'type_mappings'
+        if output.endswith("type_mappings"):
+            extension = "type_mappings"
             break
 
         # Capture two extensions if present. For example: file.javac.jar should
@@ -185,26 +185,26 @@ def GetExtension(target, extra_patterns):
         extension = ext2 + ext1  # Preserve the order in the file name.
 
         if len(extension) == 0:
-            extension = '(no extension found)'
+            extension = "(no extension found)"
 
-        if ext1 in ['.pdb', '.dll', '.exe']:
-            extension = 'PEFile (linking)'
+        if ext1 in [".pdb", ".dll", ".exe"]:
+            extension = "PEFile (linking)"
             # Make sure that .dll and .exe are grouped together and that the
             # .dll.lib files don't cause these to be listed as libraries
             break
-        if ext1 in ['.so', '.TOC']:
-            extension = '.so (linking)'
+        if ext1 in [".so", ".TOC"]:
+            extension = ".so (linking)"
             # Attempt to identify linking, avoid identifying as '.TOC'
             break
         # Make sure .obj files don't get categorized as mojo files
-        if ext1 in ['.obj', '.o']:
+        if ext1 in [".obj", ".o"]:
             break
         # Jars are the canonical output of java targets.
-        if ext1 == '.jar':
+        if ext1 == ".jar":
             break
         # Normalize all mojo related outputs to 'mojo'.
-        if output.count('.mojom') > 0:
-            extension = 'mojo'
+        if output.count(".mojom") > 0:
+            extension = "mojo"
             break
     return extension
 
@@ -229,8 +229,8 @@ def SummarizeEntries(entries, extra_step_types, elapsed_time_sorting):
         if target.end > latest:
             latest = target.end
         total_cpu_time += target.Duration()
-        task_start_stop_times.append((target.start, 'start', target))
-        task_start_stop_times.append((target.end, 'stop', target))
+        task_start_stop_times.append((target.start, "start", target))
+        task_start_stop_times.append((target.end, "stop", target))
     length = latest - earliest
     weighted_total = 0.0
 
@@ -256,35 +256,38 @@ def SummarizeEntries(entries, extra_step_types, elapsed_time_sorting):
         if num_running > 0:
             # Update the total weighted time up to this moment.
             last_weighted_time += (time - last_time) / float(num_running)
-        if action_name == 'start':
+        if action_name == "start":
             # Record the total weighted task time when this task starts.
             running_tasks[target] = last_weighted_time
-        if action_name == 'stop':
+        if action_name == "stop":
             # Record the change in the total weighted task time while this task
             # ran.
             weighted_duration = last_weighted_time - running_tasks[target]
             target.SetWeightedDuration(weighted_duration)
             weighted_total += weighted_duration
             del running_tasks[target]
         last_time = time
-    assert (len(running_tasks) == 0)
+    assert len(running_tasks) == 0
 
     # Warn if the sum of weighted times is off by more than half a second.
     if abs(length - weighted_total) > 500:
-        print('Warning: Possible corrupt ninja log, results may be '
-              'untrustworthy. Length = %.3f, weighted total = %.3f' %
-              (length, weighted_total))
+        print(
+            "Warning: Possible corrupt ninja log, results may be "
+            "untrustworthy. Length = %.3f, weighted total = %.3f"
+            % (length, weighted_total)
+        )
 
     # Print the slowest build steps:
-    print('    Longest build steps:')
+    print("    Longest build steps:")
     if elapsed_time_sorting:
         entries.sort(key=lambda x: x.Duration())
     else:
         entries.sort(key=lambda x: x.WeightedDuration())
     for target in entries[-long_count:]:
-        print('      %8.1f weighted s to build %s (%.1f s elapsed time)' %
-              (target.WeightedDuration(), target.DescribeTargets(),
-               target.Duration()))
+        print(
+            "      %8.1f weighted s to build %s (%.1f s elapsed time)"
+            % (target.WeightedDuration(), target.DescribeTargets(), target.Duration())
+        )
 
     # Sum up the time by file extension/type of the output file
     count_by_ext = {}
@@ -293,51 +296,56 @@ def SummarizeEntries(entries, extra_step_types, elapsed_time_sorting):
     # Scan through all of the targets to build up per-extension statistics.
     for target in entries:
         extension = GetExtension(target, extra_step_types)
-        time_by_ext[extension] = time_by_ext.get(extension,
-                                                 0) + target.Duration()
-        weighted_time_by_ext[extension] = weighted_time_by_ext.get(
-            extension, 0) + target.WeightedDuration()
+        time_by_ext[extension] = time_by_ext.get(extension, 0) + target.Duration()
+        weighted_time_by_ext[extension] = (
+            weighted_time_by_ext.get(extension, 0) + target.WeightedDuration()
+        )
         count_by_ext[extension] = count_by_ext.get(extension, 0) + 1
 
-    print('    Time by build-step type:')
+    print("    Time by build-step type:")
     # Copy to a list with extension name and total time swapped, to (time, ext)
     if elapsed_time_sorting:
-        weighted_time_by_ext_sorted = sorted(
-            (y, x) for (x, y) in time_by_ext.items())
+        weighted_time_by_ext_sorted = sorted((y, x) for (x, y) in time_by_ext.items())
     else:
         weighted_time_by_ext_sorted = sorted(
-            (y, x) for (x, y) in weighted_time_by_ext.items())
+            (y, x) for (x, y) in weighted_time_by_ext.items()
+        )
     # Print the slowest build target types:
     for time, extension in weighted_time_by_ext_sorted[-long_ext_count:]:
         print(
-            '      %8.1f s weighted time to generate %d %s files '
-            '(%1.1f s elapsed time sum)' %
-            (time, count_by_ext[extension], extension, time_by_ext[extension]))
+            "      %8.1f s weighted time to generate %d %s files "
+            "(%1.1f s elapsed time sum)"
+            % (time, count_by_ext[extension], extension, time_by_ext[extension])
+        )
 
-    print('    %.1f s weighted time (%.1f s elapsed time sum, %1.1fx '
-          'parallelism)' %
-          (length, total_cpu_time, total_cpu_time * 1.0 / length))
-    print('    %d build steps completed, average of %1.2f/s' %
-          (len(entries), len(entries) / (length)))
+    print(
+        "    %.1f s weighted time (%.1f s elapsed time sum, %1.1fx "
+        "parallelism)" % (length, total_cpu_time, total_cpu_time * 1.0 / length)
+    )
+    print(
+        "    %d build steps completed, average of %1.2f/s"
+        % (len(entries), len(entries) / (length))
+    )
 
 
 def main():
-    log_file = '.ninja_log'
-    metrics_file = 'siso_metrics.json'
+    log_file = ".ninja_log"
+    metrics_file = "siso_metrics.json"
     parser = argparse.ArgumentParser()
-    parser.add_argument('-C', dest='build_directory', help='Build directory.')
+    parser.add_argument("-C", dest="build_directory", help="Build directory.")
     parser.add_argument(
-        '-s',
-        '--step-types',
-        help='semicolon separated fnmatch patterns for build-step grouping')
+        "-s",
+        "--step-types",
+        help="semicolon separated fnmatch patterns for build-step grouping",
+    )
     parser.add_argument(
-        '-e',
-        '--elapsed_time_sorting',
+        "-e",
+        "--elapsed_time_sorting",
         default=False,
-        action='store_true',
-        help='Sort output by elapsed time instead of weighted time')
-    parser.add_argument('--log-file',
-                        help="specific ninja log file to analyze.")
+        action="store_true",
+        help="Sort output by elapsed time instead of weighted time",
+    )
+    parser.add_argument("--log-file", help="specific ninja log file to analyze.")
     args, _extra_args = parser.parse_known_args()
     if args.build_directory:
         log_file = os.path.join(args.build_directory, log_file)
@@ -348,34 +356,35 @@ def main():
         # Offer a convenient way to add extra step types automatically,
         # including when this script is run by autoninja. get() returns None if
         # the variable isn't set.
-        args.step_types = os.environ.get('chromium_step_types')
+        args.step_types = os.environ.get("chromium_step_types")
     if args.step_types:
         # Make room for the extra build types.
         global long_ext_count
-        long_ext_count += len(args.step_types.split(';'))
+        long_ext_count += len(args.step_types.split(";"))
 
     if os.path.exists(metrics_file):
         # Automatically handle summarizing siso builds.
-        cmd = ['siso.bat' if 'win32' in sys.platform else 'siso']
-        cmd.extend(['metrics', 'summary'])
+        cmd = ["siso.bat" if "win32" in sys.platform else "siso"]
+        cmd.extend(["metrics", "summary"])
         if args.build_directory:
-            cmd.extend(['-C', args.build_directory])
+            cmd.extend(["-C", args.build_directory])
         if args.step_types:
-            cmd.extend(['--step_types', args.step_types])
+            cmd.extend(["--step_types", args.step_types])
         if args.elapsed_time_sorting:
-            cmd.append('--elapsed_time_sorting')
+            cmd.append("--elapsed_time_sorting")
         subprocess.run(cmd)
     else:
         try:
-            with open(log_file, 'r') as log:
+            with open(log_file, "r") as log:
                 entries = ReadTargets(log, False)
                 if entries:
-                    SummarizeEntries(entries, args.step_types,
-                                     args.elapsed_time_sorting)
+                    SummarizeEntries(
+                        entries, args.step_types, args.elapsed_time_sorting
+                    )
         except IOError:
-            print('Log file %r not found, no build summary created.' % log_file)
+            print("Log file %r not found, no build summary created." % log_file)
             return errno.ENOENT
 
 
-if __name__ == '__main__':
+if __name__ == "__main__":
     sys.exit(main())

ci/sccache_hit_rate.sh

@@ -37,5 +37,5 @@ if [ $requests_diff -eq 0 ]; then
 else
     hit_rate=$(awk -v hits=$hits_diff -v requests=$requests_diff 'BEGIN {printf "%.2f", hits/requests * 100}')
     echo "sccache hit rate: $hit_rate%" >&2
-    echo "$hit_rate" 
+    echo "$hit_rate"
 fi

cmake/DetectSupportedStandards.cmake

@@ -1,4 +1,4 @@
-# Detect the langauge standards supported by the current compilers.
+# Detect the language standards supported by the current compilers.
 #
 # Usage: detect_supported_cxx_standards(<var_prefix> <lang> <standards>)
 #