v3.0.0

Key changes:

• Remove the generator prefix from the CLI and code as that was premature generalization.
• Improve mid-range (e.g. 100K files) performance by ~10%.

v2.0.3

Signed-off-by: Alex Saveau <saveau.alexandre@gmail.com>

v2.0.2

Signed-off-by: Alex Saveau <saveau.alexandre@gmail.com>

v2.0.1

Signed-off-by: Alex Saveau <saveau.alexandre@gmail.com>

v2.0.0

The quality of directory shapes was dramatically improved.

Results

• Per directory file count follows a much broader and non-normal distribution resulting in more varied directories.
• Total file count has lasered into the true mean instead of skewing upwards proportional to max depth (atypical parameters still result in skew, but it is at least a normal-ish distribution).

‌	Old	New
Files per directory
Total file count

Total file count	Old	New
Low file count
High depth

Appendix

Sample collection code:

Subject: [PATCH] Upgrade deps
---
Index: src/core/scheduler.rs
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/src/core/scheduler.rs b/src/core/scheduler.rs
--- a/src/core/scheduler.rs	(revision 6b7e88c4d05ed8fd247178bd9ec86440fa923d5f)
+++ b/src/core/scheduler.rs	(date 1679085314728)
@@ -11,6 +11,7 @@
     core::{
         files::GeneratorTaskOutcome,
         tasks::{QueueErrors, QueueOutcome, TaskGenerator},
+        TOTAL_FILE_COUNT,
     },
     generator::Error,
     utils::{with_dir_name, FastPathBuf},
@@ -165,6 +166,11 @@
         handle_task_result(task, &mut stats)?;
     }
 
+    {
+        use std::io::Write;
+        let mut count = TOTAL_FILE_COUNT.lock().unwrap();
+        writeln!(count, "{}", stats.files).unwrap();
+    }
     Ok(stats)
 }
 
Index: src/core/mod.rs
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/src/core/mod.rs b/src/core/mod.rs
--- a/src/core/mod.rs	(revision 6b7e88c4d05ed8fd247178bd9ec86440fa923d5f)
+++ b/src/core/mod.rs	(date 1679085426452)
@@ -1,3 +1,9 @@
+use std::{
+    cell::LazyCell,
+    fs::File,
+    sync::{LazyLock, Mutex},
+};
+
 pub use scheduler::{run, GeneratorStats};
 pub use tasks::{DynamicGenerator, GeneratorBytes, StaticGenerator};
 
@@ -5,3 +11,31 @@
 mod files;
 mod scheduler;
 mod tasks;
+
+static FILE_COUNT: LazyLock<Mutex<File>, fn() -> Mutex<File>> = LazyLock::new(|| {
+    Mutex::new(
+        File::options()
+            .create(true)
+            .append(true)
+            .open("file_count.samples")
+            .unwrap(),
+    )
+});
+static DIR_COUNT: LazyLock<Mutex<File>, fn() -> Mutex<File>> = LazyLock::new(|| {
+    Mutex::new(
+        File::options()
+            .create(true)
+            .append(true)
+            .open("dir_count.samples")
+            .unwrap(),
+    )
+});
+static TOTAL_FILE_COUNT: LazyLock<Mutex<File>, fn() -> Mutex<File>> = LazyLock::new(|| {
+    Mutex::new(
+        File::options()
+            .create(true)
+            .append(true)
+            .open("total_file_count.samples")
+            .unwrap(),
+    )
+});
Index: src/lib.rs
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/src/lib.rs b/src/lib.rs
--- a/src/lib.rs	(revision 6b7e88c4d05ed8fd247178bd9ec86440fa923d5f)
+++ b/src/lib.rs	(date 1679084880740)
@@ -3,6 +3,7 @@
 #![feature(let_chains)]
 #![feature(const_option)]
 #![feature(inline_const)]
+#![feature(once_cell)]
 #![allow(clippy::multiple_crate_versions)]
 #![allow(clippy::module_name_repetitions)]
 
Index: src/core/tasks.rs
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/src/core/tasks.rs b/src/core/tasks.rs
--- a/src/core/tasks.rs	(revision 6b7e88c4d05ed8fd247178bd9ec86440fa923d5f)
+++ b/src/core/tasks.rs	(date 1679085314852)
@@ -12,6 +12,7 @@
             PreDefinedGeneratedFileContents,
         },
         files::{create_files_and_dirs, GeneratorTaskOutcome, GeneratorTaskParams},
+        DIR_COUNT, FILE_COUNT,
     },
     utils::FastPathBuf,
 };
@@ -122,6 +123,16 @@
         } else {
             0
         };
+        {
+            use std::io::Write;
+            let mut count = FILE_COUNT.lock().unwrap();
+            writeln!(count, "{}", num_files).unwrap();
+        }
+        {
+            use std::io::Write;
+            let mut count = DIR_COUNT.lock().unwrap();
+            writeln!(count, "{}", num_dirs).unwrap();
+        }
 
         macro_rules! build_params {
             ($file_contents:expr) => {{

Collection:

cargo b --features dry_run
for d in [0, 5, 20] { for n in [10, 10_000, 1_000_000] { for i in 0..1000 { ./target/debug/ftzz g -n $n /tmp/foo -d $d --seed $i out> /dev/null }; mkdir $"($d)d_($n)n"; mv *.samples $"($d)d_($n)n/" } }

Plots:

!unzip samples.zip
import numpy as np
import matplotlib.pyplot as plt
import os

for dir in os.listdir():
  if not dir.endswith('n'):
    continue

  # List of files containing samples
  files = ['file_count.samples', 'total_file_count.samples']

  # Loop over each file and plot its samples

  for file in files:
      # Load samples from file into a list
      file = dir + '/' + file
      with open(file, 'r') as f:
          samples = [int(x.strip()) for x in f.readlines()]

      # Calculate some statistics on the samples
      mean = np.mean(samples)
      std = np.std(samples)

      # Plot the histogram of the samples
      plt.hist(samples, bins=50, density=True, alpha=0.5)

      # Add a vertical line at the mean of the distribution
      plt.axvline(x=mean, color='red', linestyle='--')

      # Add labels and title to the plot
      plt.xlabel('Values')
      plt.ylabel('Frequency')
      plt.title(f'Distribution of samples from {file}')

      # Show the plot
      plt.show()

v1.2.0

• Added the fill-byte flag which lets you fill files with a specific byte instead of pseudo-random ones.

v1.1.4

• Slight efficiency improvements
• Removed some unnecessary unsafe + other cleanup

v1.1.3

Tweak help output to be slightly clearer.

v1.1.2

Re-release of 1.1.1 with fixed binary builds. Binaries are now statically linked.

v1.1.1

Improvements across internal infrastructure, testing, and error handling.