SyzygyCheck v2.2.0

SyzygyCheck v2.2.0 — Controlemap and subfolder scanning

v2.2.0 follows v2.1.0 and the successfully validated v2.0.9 V9F release. The verification/checksum core,
thread functionality and compact live progress renderer are unchanged. V9F remains available untouched as the
first practically validated release.

v2.2.0 changes

• Before inventory, the user chooses:
  1 = Controlemap only [default]
  2 = Controlemap and all subfolders
• A run never scans a parent folder.
• Real subfolders are followed to any depth; symbolic links are excluded to preserve the hard boundary.
• Whole-drive scans skip $RECYCLE.BIN and System Volume Information, preventing deleted or protected tablebase
  filenames in Windows metadata from aborting the real scan.
• Relative subfolder paths are preserved in checkpoint state for safe resume and duplicate filename handling.
• Every FAIL and ERROR in the final report shows its full Windows path from the drive letter.
• Immediately before saving the final report, the user chooses:
  1 = Desktop [default on first use]
  2 = Controlemap
  3 = freely selected folder through a native Windows Explorer-style folder dialog
• SyzygyCheck automatically creates a result collection subfolder inside the selected location.
• The last selected destination becomes the Enter-default for the next run. A freely selected path is remembered
  exactly; if it later disappears or is unavailable, a seven-language warning shows the previous path and states
  that Desktop is now the default. Pressing Enter then saves to Desktop and replaces the invalid preference.
• The new scope and destination menus are resize-safe, including long remembered paths and CJK console-cell widths.
• That folder is named CheckResult in English, German, Dutch, French and Spanish; Chinese and Russian use a
  translated name in their own script.
• Consecutive reports retain their timestamped filenames and collect in that folder instead of being scattered
  across the Desktop.
• The executable filename inside the release ZIP always begins with an exclamation mark.
• The final percentage/remaining-time block is followed by exactly one blank line before the report-destination
  prompt, keeping the two interface sections visually separate.
• The complete user ZIP now contains the required four-part structure:
  !SyzygyCheck_v2.2.0.exe
  1_READ_FIRST_AL.txt
  Docs_AL
  Program_AL\
• FAIL/ERROR reports show the exact packaged recovery path under Program_AL.
• Source is not included in the user ZIP. The recommended source download is the clearly named
  SyzygyCheck_v2.2.0_SOURCE.zip release asset; GitHub's automatically generated tag archives remain available too.

Retained V9F refinements

• Main-menu prompt simplified in all seven languages: only the Enter default is shown. The multilingual
  Change language menu item already makes language switching obvious, so Esc = language is no longer
  printed on the main-menu prompt.
• Russian remaining times of at least one hour now use ч (h).
• Chinese remaining times of at least one hour now use 小时 (h).
• English, German, Dutch, French and Spanish continue to use the compact international h.
• Release documentation now distinguishes classic-console close protection from Windows Terminal behaviour.

Verification core

Based on the original Syzygy tablebase verification code by Ronald de Man.
The AL layer does not reimplement the checksum logic; it adds the multilingual interface, progress display,
self-test, checkpoint/resume, thread selection and reporting around the original verification core.

Thread selection

At startup SyzygyCheck silently detects the logical processors available to the process and offers:

1 = all detected threads
2 = half [default; Enter selects this]
3 = a custom value from 1 to the detected maximum

The selected count is visible in the run header using the properly localised equivalent of Workers (threads)
and is passed directly to tbcheck as --threads n.

Live display

• compact three-line live progress block;
• read speed on the same first line as the current file count;
• read speed and remaining time share one right-hand column when width permits;
• exactly one blank separator after a visible FAIL/ERROR;
• current progress remains visible while older error lines can scroll upward;
• checkpoint/resume, onboard intentional FAIL self-test and reporting are retained.

Escape/back navigation

Physical Esc remains a genuine back action in Windows choice screens. On the main menu its language shortcut
still works but is no longer advertised because the visible language-change menu item is already sufficient.

Window-close protection

During the real scan SyzygyCheck still makes a best-effort attempt to disable SC_CLOSE for a classic Windows
console host. This does not control the outer top-right X of Windows Terminal, because that button belongs to
Windows Terminal itself. In that host an accidental close can therefore still terminate the visible terminal.
Checkpoint/resume remains the dependable protection against losing already confirmed work.

Minimize, maximize, resizing and scrollback are not intentionally blocked.

Read-speed meaning

The displayed value is effective verification throughput, not necessarily raw physical disk speed. Confirmed
results and the final report use definitively verified bytes for the current session; bytes completed before a
resumed session are excluded from that session average.

Checkpoint behaviour

Confirmed FAIL results are written into the checkpoint and are carried into a resumed run. A partially completed
batch may be rechecked. A fully completed run without operational ERROR removes the checkpoint.

Final practical validation

• V9F, i7-6800K, 11 workers: 57 files / 20.22 GB, 27 s, 777.4 MB/s average effective verification throughput,
  exactly 2 expected checksum FAILs found, 0 read/process errors, onboard diagnostic TEST-FAIL passed.
• Earlier cold V9C/V9D/V9E-line validation on the same i7-6800K with 12 workers reached 805.0 MB/s average
  effective verification throughput after reboot.
• Drobo: successful practical operation with 4 threads, acceptable throughput and reasonably normal mechanical
  behaviour.
• Large X299 / 5× NVMe RAID0 scaling remains useful as an additional performance measurement, but is not a
  release blocker.

Executable

!SyzygyCheck_v2.2.0.exe

Release gate

Before publication the workflow runs Go tests and vet, builds Windows x64, assembles the complete user ZIP, checks
every required file and directory, rejects Go source inside the user ZIP, checks all seven exact recovery paths,
and creates a SHA-256 file for the final release ZIP. Publication occurs only after this gate succeeds.

SyzygyCheck v2.1.0

SyzygyCheck v2.1.0 — report destination selection

v2.1.0 follows the successfully validated v2.0.9 V9F release. The verification/checksum core, thread
functionality, checkpoint logic and compact live progress renderer are unchanged. V9F remains available
untouched as the first practically validated release.

v2.1.0 changes

• Immediately before saving the final report, the user chooses:
  1 = Desktop [default]
  2 = checked folder
  3 = freely selected folder through a native Windows Explorer-style folder dialog
• SyzygyCheck automatically creates a result collection subfolder inside the selected location.
• That folder is named CheckResult in English, German, Dutch, French and Spanish; Chinese and Russian use a
  translated name in their own script.
• Consecutive reports retain their timestamped filenames and collect in that folder instead of being scattered
  across the Desktop.
• The executable filename now always begins with an exclamation mark.

Retained V9F refinements

• Main-menu prompt simplified in all seven languages: only the Enter default is shown. The multilingual
  Change language menu item already makes language switching obvious, so Esc = language is no longer
  printed on the main-menu prompt.
• Russian remaining times of at least one hour now use ч (h).
• Chinese remaining times of at least one hour now use 小时 (h).
• English, German, Dutch, French and Spanish continue to use the compact international h.
• Release documentation now distinguishes classic-console close protection from Windows Terminal behaviour.

Verification core

Based on the original Syzygy tablebase verification code by Ronald de Man.
The AL layer does not reimplement the checksum logic; it adds the multilingual interface, progress display,
self-test, checkpoint/resume, thread selection and reporting around the original verification core.

Thread selection

At startup SyzygyCheck silently detects the logical processors available to the process and offers:

1 = all detected threads
2 = half [default; Enter selects this]
3 = a custom value from 1 to the detected maximum

The selected count is visible in the run header using the properly localised equivalent of Workers (threads)
and is passed directly to tbcheck as --threads n.

Live display

• compact three-line live progress block;
• read speed on the same first line as the current file count;
• read speed and remaining time share one right-hand column when width permits;
• exactly one blank separator after a visible FAIL/ERROR;
• current progress remains visible while older error lines can scroll upward;
• checkpoint/resume, onboard intentional FAIL self-test and reporting are retained.

Escape/back navigation

Physical Esc remains a genuine back action in Windows choice screens. On the main menu its language shortcut
still works but is no longer advertised because the visible language-change menu item is already sufficient.

Window-close protection

During the real scan SyzygyCheck still makes a best-effort attempt to disable SC_CLOSE for a classic Windows
console host. This does not control the outer top-right X of Windows Terminal, because that button belongs to
Windows Terminal itself. In that host an accidental close can therefore still terminate the visible terminal.
Checkpoint/resume remains the dependable protection against losing already confirmed work.

Minimize, maximize, resizing and scrollback are not intentionally blocked.

Read-speed meaning

The displayed value is effective verification throughput, not necessarily raw physical disk speed. Confirmed
results and the final report use definitively verified bytes for the current session; bytes completed before a
resumed session are excluded from that session average.

Checkpoint behaviour

Confirmed FAIL results are written into the checkpoint and are carried into a resumed run. A partially completed
batch may be rechecked. A fully completed run without operational ERROR removes the checkpoint.

Final practical validation

• V9F, i7-6800K, 11 workers: 57 files / 20.22 GB, 27 s, 777.4 MB/s average effective verification throughput,
  exactly 2 expected checksum FAILs found, 0 read/process errors, onboard diagnostic TEST-FAIL passed.
• Earlier cold V9C/V9D/V9E-line validation on the same i7-6800K with 12 workers reached 805.0 MB/s average
  effective verification throughput after reboot.
• Drobo: successful practical operation with 4 threads, acceptable throughput and reasonably normal mechanical
  behaviour.
• Large X299 / 5× NVMe RAID0 scaling remains useful as an additional performance measurement, but is not a
  release blocker.

Executable

!SyzygyCheck_v2.1.0.exe

SyzygyCheck v2.0.9 V9F — Final

SyzygyCheck v2.0.9 V9F — Final

Multilingual Windows integrity checker for Syzygy chess tablebases.

Based on the original Syzygy tablebase verification code by Ronald de Man.
The AL layer adds the multilingual interface, progress display, self-test,
checkpoint/resume, thread selection and reporting around the original
verification core.

Highlights:

• English, German, Dutch, French, Spanish, Chinese and Russian
• Automatic logical-processor detection
• Thread selection: all / half [default] / custom
• Checkpoint and resume support
• Confirmed checksum failures are preserved when resuming
• Live effective verification speed and remaining time
• Detailed final report
• Diagnostic onboard self-test
• Recovery and pagefile guidance in multiple languages

Final practical validation:

• 57/57 files checked
• 20.22 GB
• 11 workers (threads)
• 27 seconds
• 777.4 MB/s average effective verification throughput
• exactly 2 intentional checksum FAILs detected
• 0 read/process errors
• onboard diagnostic TEST-FAIL passed

Windows Terminal note:
The outer Close (X) button belongs to Windows Terminal itself and cannot
reliably be disabled by SyzygyCheck. Checkpoint/resume protects already
confirmed progress after an accidental close.

License: GPL-2.0