FileAtlas 2.0.0

[File-Maker]

FileAtlas 2.0.0 Release Description

FileAtlas 2.0.0 is the production Core release for Base, FAFS, FAStore, Analytics, History, global search, Dev diagnostics, licensing compatibility, and the benchmark/proving path.

This release documentation replaces the earlier draft documentation that described the Core work as an upcoming 2.1.3 source release while keeping 2.0.0 executables. The product being released is FileAtlas 2.0.0.

The main release boundary is now explicit:

• FileAtlas Core 2.0.0 is supported and ready for normal use/pilot deployment.
• FAAV is not being developed, validated, or supported for this version. Do not install FAAV for normal use.

Older FAAV binaries, installer options, owner tools, or source paths may still exist. Their presence does not make them part of the supported 2.0.0 release. If someone deliberately installs/runs them anyway, that use is entirely unsupported and at the user's own risk.

Download

Current installer set:

FileAtlasSetup-2.0.0.exe
FileAtlasLicenseManagerSetup-2.0.0.exe  (owner-only)
FileAtlasBenchmarkSetup-2.0.0.exe   (owner-only)
FileAtlasFAAVLabSetup-2.0.0.exe   (owner-only legacy/experimental FAAV tooling; not a supported public component)

Current SHA-256 values:

9a76586d6e40222529ca4c26a2b2c88f07a439ae158facc9aea0a94cc166d726  installer/FileAtlasSetup-2.0.0.exe
e5af82fb1c44234c952c13b4d580c1d942b089d84b3a7b9c0a9fa4e10015b27d  installer/FileAtlasLicenseManagerSetup-2.0.0.exe
83cceffdac721090c8782bc6f397afa683ac5689e1e2e1a0bfd11b00e985cbfa  installer/FileAtlasBenchmarkSetup-2.0.0.exe
f9641ece77e6931336d6278cd4036b9b760a23a6b6679ea2c6bcb972ce0b52bc  installer/FileAtlasFAAVLabSetup-2.0.0.exe

SHA256SUMS.txt is authoritative for the supplied installer set.

What 2.0.0 contains

1. Canonical Core truth architecture

FAFS remains the authoritative catalog, mutation, continuity, and history service. Base can contribute learned directory information through an acknowledged handoff, but it does not become a second competing catalog authority.

Core state now keeps identity, path, time, capture source/quality, and current/last-known measurement semantics separate enough to prevent older evidence from silently overwriting newer truth.

2. Volume-wide live filesystem history

On NTFS, FAFS consumes the volume-wide USN change journal rather than only watching the directory currently open in Base.

This means FileAtlas can retain activity across the volume, including:

• creates;
• edits;
• deletes;
• replacements;
• moves;
• renames;
• Recycle Bin movement;
• folder activity.

A current Base folder is a priority hint, not the boundary of what FileAtlas can observe.

3. Restart continuity and independent volume state

FAFS uses saved journal cursors for incremental restart catch-up when journal continuity remains valid.

When continuity is lost or a volume requires a baseline/reconciliation pass, FileAtlas does not pretend that nothing happened. It separates:

• live ingestion;
• historical replay/catch-up;
• baseline/reconciliation;
• aggregation;
• recovery/parity metadata work.

Volumes are handled independently. A noisy or recovering C:\ volume does not make a healthy D:\ volume stop being current.

The final Core corrective work also addressed state-convergence problems where individual lanes could already be complete/current while a higher-level volume state remained stuck in CATCHING_UP.

4. Exact identity-led History

Analytics contains a dedicated History view for folder activity and exact item history.

The important 2.0.0 behavior is that item history follows object identity, not merely whatever path/name currently looks similar.

The Core corrective pass hardened this area after several real failure cases:

• unrelated history could be returned when exact identity resolution failed;
• one failure mode could make different Item History requests resolve to the same unrelated object;
• parent-directory renames could leave a child trail showing a stale projected present at path;
• deleted/recycled items needed to end in a correct absent state rather than inheriting stale current-path state.

The corrected behavior is fail-closed: legacy records that do not contain enough identity evidence are excluded from exact item history instead of being guessed into the requested object's trail.

History can now show lifecycle sequences such as:

edited
→ moved to Recycle Bin
→ removed from Recycle Bin
→ no longer present

while preserving the same native identity when that evidence is available.

5. Parent-rename and nested-lineage continuity

History persists native-identity postings in addition to path-oriented evidence.

That allows a file/folder trail to survive cases where:

• the item itself is renamed;
• the item is moved;
• a parent directory is renamed;
• the item is later recycled or deleted.

The key design rule is that a path is an attribute of an object at a point in time; it is not the object's identity.

6. Search

FileAtlas Base keeps the existing current-directory filter and adds a separate Search all path backed by FAFS catalog truth.

Global search can use exact names, paths, normalized tokens, prefixes, and bounded partial-name matching across known files and folders.

Opening a global result projects the parent into Base and focuses the requested file/folder rather than starting an unrelated deep scan.

Analytics History also has its own search/filter controls so retained history can be narrowed by name/path and event type.

7. Deleted-object discovery

A deleted object can still be found in retained history after it is no longer present on disk.

The final Core proof demonstrated a history probe whose current object no longer existed while FileAtlas still showed the exact retained lifecycle and correctly reported:

no longer present

This is one of the central product distinctions between a current-state disk visualizer and a filesystem historian.

8. C:\ real-world history path

The final corrective proof was not limited to a quiet secondary drive.

C:\ successfully returned real high-volume history including Recycle Bin/system-path activity and exact event details. This was important because C:\ had previously been the volume most likely to expose:

• journal discontinuity/rebaseline paths;
• memory/resource pressure;
• mixed recovery/catch-up state;
• high system/background churn.

The final Core build demonstrated that the history path remains usable on that real workload rather than only in isolated test folders.

9. Base and Analytics behavior

Base:

• zoomable/pannable treemap;
• current-folder filter;
• global Search all;
• size filters;
• cleanup queue;
• Recycle Bin deletion where supported;
• duplicate marking from file-content hashes where available;
• single UI instance per Windows session;
• selected-item History action;
• last-known completed views while FAFS reconciles.

Analytics:

• added/cleaned/net storage flow;
• change counts and Recycle Bin activity;
• drive/folder drill-downs;
• recent activity;
• exact time ranges;
• dedicated folder/item History;
• current/as-of presence reporting;
• search/filter controls;
• last valid exact-range result retained while FAFS is temporarily unavailable/restarting.

10. Local-only architecture

FileAtlas Core remains local-only.

It does not upload:

• files;
• filenames;
• scan results;
• history;
• analytics;
• hashes;
• settings;
• licenses;
• benchmark results;
• debug reports.

FAFS exposes local-only services on 127.0.0.1 for same-machine FileAtlas components.

11. Dev, LicenseManager, and Benchmark

FileAtlas Dev remains an optional diagnostic application. Privileged features require a local .falicense validated by FAFS.

LicenseManager remains an owner-only license issuing/inspection tool.

Benchmark remains a separate local stress-test/report application using isolated dummy data.

Final installed-build proof

The final Core corrective pass included a focused proof on the installed application rather than only source/unit tests.

Observed on the release machine:

• installation completed successfully;
• installed FAFS binary matched the built artifact exactly;
• an existing catalog of approximately 1.8 million items was preserved/reopened;
• FAFS reported no watcher errors in the focused proof;
• a create became visible in about 182 ms in that run;
• tested deletions became visible in about 33 ms in that run;
• file history showed added/edited/removed lifecycle state using the same native identity;
• exact history correctly ended in absent / no longer present state;
• real C:\ history and Recycle Bin activity rendered successfully;
• exact item-history resolution no longer substituted the unrelated stale identity seen in the earlier failure;
• identity-less legacy rows were disclosed/excluded rather than silently merged into an exact object trail.

These measurements are local observations from the release machine and are not universal latency guarantees.

Important corrections from the earlier 2.1.3 draft documentation

The previous documents contained several statements that are no longer appropriate for the release being shipped.

Product version

Earlier draft wording:

FileAtlas 2.1.3

Release wording:

FileAtlas 2.0.0

There is no separate public 2.1.3 product represented by these documents.

FAAV

Earlier draft documentation described FAAV as an optional integrated protection component and listed detailed FAAV architecture/capability claims.

For this release, that is not the supported product boundary.

FAAV is not being developed or validated for FileAtlas 2.0.0 and should not be installed by normal users. Any existing FAAV binaries, component options, source trees, benchmark simulations, or FAAV Lab packages are legacy/experimental artifacts outside the supported 2.0.0 Core release.

This release therefore does not claim:

• representative real-world malware detection effectiveness;
• supported automatic quarantine behavior;
• supported FAAV recovery behavior;
• supported behavioral-model calibration;
• supported FAAV production deployment;
• independent antivirus validation.

Recovery

FAFS contains recovery-oriented metadata/parity infrastructure, but FileAtlas 2.0.0 does not ship a supported user-facing file-recovery product.

History/provenance and recovery are different features. Knowing that a deleted object existed and preserving its identity/path history does not mean FileAtlas kept a restorable copy of the file.

Universal file editing/viewing

FileAtlas Dev remains a diagnostic/internal inspection tool.

The following are not claimed as 2.0.0 features:

• universal file editing;
• PDF editing;
• Word/Office editing;
• CAD;
• Photoshop-like image editing;
• arbitrary safe round-trip editing of unknown binary/container formats.

Generic/raw inspection discovered during development should not be confused with a supported universal editor.

Code signing

The public installer is not code-signed yet. Windows SmartScreen may therefore warn about the installer.

Historical performance reference

The earlier FileAtlas 1.8.0 maximum-stress benchmark remains a useful historical reference but was not rerun as the final 2.0.0 Core corrective proof.

That historical benchmark used isolated dummy data with:

1,000,000 catalog records
500,000 timestamped history events over seven days
500,000-node treemap
50,000 dummy files

Selected historical release-machine measurements were:

Bulk catalog learning: 17,330 records per second
500,000-node treemap layout: 358,544 nodes per second
Zoom/pan viewport transforms: 3,359 operations per second
Compact history rollups (1h / 24h / 7d): 110 / 88 / 117 ms
Analytics summaries (1h / 24h / 7d): 83 / 85 / 121 ms
Populated FAStore cold reopen (1,000,000 catalog + 500,000 history): 5,188 ms
Full million-record maintenance compaction: 33,280 ms
Average benchmark-process whole-machine CPU equivalent: 5.75%
Peak synthetic harness memory: 3.56 GB

Those numbers describe that historical synthetic run and are not guarantees for other machines or the final production workload.

What is not included in the supported 2.0.0 release

For clarity, the supported release does not include or promise:

• production-supported FAAV;
• cloud malware reputation/sample upload;
• cloud sync or cloud Analytics;
• remote telemetry collection;
• a user-facing FileAtlas Recovery product;
• full backup copies of user files;
• universal safe editing of arbitrary binary/container formats;
• PDF/Office/CAD editing;
• code signing/SmartScreen reputation;
• independent antivirus certification/evaluation.

Safety and privacy

Normal FileAtlas delete actions use the Windows Recycle Bin where supported. Protected/high-risk system locations are blocked or strongly warned.

FileAtlas history may retain metadata showing that an object existed after the object itself has been deleted. That history remains local unless the user manually exports/shares something.

See PRIVACY.md for the detailed local-data policy.

Support

iakobidzesab@gmail.com

---

This discussion was created from the release FileAtlas 2.0.0.