RiptOPL Status Update — October 13, 2026

[NathanNeurotic]

Comparison baseline: October 9 weekly report
Scope: Material changes only

Regression investigations

UDPFS moved from diagnosis to concrete merged fixes

This is the most important change since October 9.

PR [#921](#921) identified and fixed three transport behaviors that made RiptOPL's UDPFS implementation less reliable than the stock Neutrino-derived implementation used successfully elsewhere:

• Rediscovery now resets TX/RX sequence state instead of carrying a dead session's sequence numbers into the new session.
• Stale ACKs no longer consume retries and trigger unnecessary retransmission/session loss.
• The PS2 sends a gratuitous ARP before UDPFS discovery, repairing cases where the server PC has a stale or conflicting ARP entry.

The old behavior directly explains the reported rediscovery loop / games never appear / first boot works then subsequent launch fails pattern.

PR [#922](#922) then hardened the surrounding network lifecycle:

• RA menu operations no longer accidentally initialize SMB while UDPFS/UDPBD is selected.
• UDP transports are gated correctly on successful DEV9 initialization.
• Failed DEV9 initialization can be retried rather than leaving an unusable partially initialized state.
• UDPFS modules are rebuilt reliably when headers change, preventing stale embedded IRXs.
• RiptOPL can now detect another MAC using the configured PS2 IP and warn about the conflict.

This materially strengthens the earlier conclusion that the observed failure was RiptOPL-side, rather than a PS2-Servers problem.

PS2-Servers still documents explicit data-interface binding as necessary on some multi-interface PCs, so host-side routing remains a separate possible failure mode rather than being confused with the loader transport regression.

Remaining blocker: the fixes are extensively host-tested, but the critical UDPFS regression still needs confirmation on the affected physical-console/server combinations.

---

RetroAchievements / parity work

This area advanced substantially.

Consolidated RA/Caduceus work merged

PR [#894](#894), which was still outstanding in the October 9 report, merged October 10.

That closes the major integration line covering USB PS1 achievements, Caduceus support, and related integration hardening.

ORBIT parity became production work

PR [#916](#916) merged October 12, bringing ORBIT-derived parity work into the RA build and establishing the dedicated RA release channel.

A key policy was also hardened: RetroAchievements must never prevent a game from launching. If tracking cannot be initialized, RiptOPL launches the game untracked rather than treating RA as a launch requirement.

Neutrino RA parity improved

PR [#923](#923) merged October 13.

RiptOPL now follows ORBIT behavior for achievement sets containing pointer chains: supported direct memory entries remain tracked rather than rejecting the entire game because some achievements require chain support.

Caduceus-mode Neutrino launches now also participate in the same telemetry path where possible.

UDPFS is no longer excluded from RA tooling

PRs [#927](#927) and [#929](#929) merged October 13.

The UDPFS ministack now allows:

• Check game support while a UDPFS game is selected.
• Caduceus achievement browsing over UDPFS Files.
• RA communication without bringing up a competing TCP/IP/SMB network stack.
• Correct handling of the UDPFS driver's actual configured PS2 address.
• Clearer messages when the UDPFS chain has not yet been initialized.
• ART/ and art/ key-directory compatibility on case-sensitive servers.

This removes a significant parity limitation caused by UDPFS owning the network adapter.

Online game-ID lookup merged

PR [#930](#930) merged today.

RA builds can now perform an HTTPS hash-to-GameID lookup against RetroAchievements, allowing the achievement browser to identify games that have never previously been played on the user's account.

The implementation is RA-build-only and uses mbedTLS with certificate and hostname validation.

Two important unrelated bugs were uncovered during this work:

1. RiptOPL's vendored ps2ips DNS code copied the address-list pointer instead of the returned IP address bytes.
2. The network timeout calculation overflowed because the PS2 clock counter was being treated incorrectly.

Both are now fixed.

The public RetroAchievements repositories remain actively maintained, but no external RAWeb/rcheevos change found in this check currently supersedes RiptOPL's new integration work.

Major caveat: #927, #929 and #930 explicitly remain ZERO HW implementations. Real EE TLS handshake cost, UDPFS ministack behavior, and failure/retry behavior require console validation.

---

Other material changes

PR [#917](#917) merged UTF-8 Japanese/CJK title rendering support without replacing the normal UI font.

PR [#931](#931) fixed the Rolling Release pipeline being able to stall indefinitely on a MEGA mirror upload. MEGA is now bounded to ten minutes and cannot prevent GitHub release normalization or the RA release from completing.

This was a real release-channel failure: Beta-3911 had been published as loose raw assets while the packaging/normalization stage remained blocked.

PR [#932](#932) is currently open, adding optional CFG Title= values to game lists.

The feature is off by default and preserves underlying filename identity, but enabling it requires reading each game's CFG during list rebuilding. Large SMB/MMCE/MX4SIO libraries therefore need real performance testing before merge.

---

Key decisions

Several architecture decisions are now clearer than they were on October 9:

• RA is optional telemetry, never a game-launch dependency.
• UDPFS must retain ownership of its network environment rather than allowing helper features to opportunistically initialize SMB.
• UDPFS recovery should repair the actual protocol/session state instead of repeatedly rediscovering around stale state.
• ORBIT behavior is being treated as the parity reference where its runtime behavior is demonstrably compatible with xeRAbora.
• RA-only dependencies such as mbedTLS remain isolated from ordinary RiptOPL builds.
• Network conflicts should be surfaced explicitly rather than appearing as unexplained empty lists or black screens.

---

Unresolved blockers

The principal blocker has shifted from implementation to hardware confirmation.

Highest-risk unverified paths are now:

1. UDPFS reconnect behavior after restart/disconnect.
2. UDPFS + Ember launches on the configurations that previously black-screened.
3. UDPFS RA ministack communication.
4. HTTPS RetroAchievements lookup performance on the Emotion Engine.
5. Caduceus browsing and session restoration across UDPFS.
6. Large-library CFG-title refresh performance if Optional CFG Titles in game lists, off by default (#918) #932 is considered for merge.

Older VMC/controller/refresh regressions remain relevant, but nothing discovered since October 9 justifies reprioritizing them above validating the newly merged UDPFS transport repair.

---

Next highest-priority actions

1. Hardware-test fix(udpfs): heal reconnects, keep waiting on stale ACKs, announce the PS2 before discovery #921/fix(network): keep UDP transports loadable, gate them on DEV9, warn on IP clashes #922 first. Reproduce the exact previously failing PS2-Servers + UDPFS scenarios, including server restart and repeated RiptOPL boot/launch cycles.
2. Test Ember over UDPFS. Launchstation and other implementations provide a useful control because UDPFS/Ember operation is known to work outside the problematic RiptOPL path.
3. Exercise RA: support check on UDPFS games + Caduceus browser under UDPFS (mstack: link) #927/RA: mstack follow-ups from the #927 review #929 on console. Verify support lookup and Caduceus browsing while UDPFS owns DEV9.
4. Benchmark RA: HTTPS game-ID lookup like ORBIT (mbedTLS, RA builds only) #930 HTTPS. Measure first lookup/handshake time on a real PS2, then test no-network, incorrect-clock and certificate-error paths.
5. Do not merge Optional CFG Titles in game lists, off by default (#918) #932 solely from host tests. Time list rebuilding against a realistically large SMB/MMCE library first.
6. Once those paths are confirmed, return to the remaining controller/VMC/automatic-refresh regressions rather than expanding the feature surface further.

Overall

The last few days were materially busier than the October 9 report.

The major UDPFS regression now has a credible technical root cause and merged repair rather than only defensive workarounds. At the same time, RiptOPL's RetroAchievements implementation moved significantly closer to ORBIT parity, including Neutrino handling, Caduceus, UDPFS network coexistence and direct HTTPS game identification.

The project's immediate risk is no longer primarily missing implementation. It is that a large amount of important networking and RA code has landed faster than it has been validated on physical hardware.

The next cycle should therefore emphasize targeted console validation over another large feature batch.