เอกสารภาษาอังกฤษ
หน้านี้ยังไม่มีฉบับภาษาไทย จึงแสดงฉบับภาษาอังกฤษพร้อมแหล่งอ้างอิง
Recovery evidence, missing dates and Episode labels
Research update: 2026-10-28. No emulator changes. The full patch history now gives every stage an Episode/era label, its source and its evidence status.
Round-five additions below supersede earlier missing-schedule and missing-guide claims. Use the timeline bible and its research gaps for the current paired guides and work queue.
Thai recovery: what the original guide actually establishes
The original operator’s leveling guide, preserved December 12, 2003, links red weight with suspended natural HP/SP recovery. It says weight over 50%; that wording does not resolve the exact >50% versus >=50% boundary. There is no natural-recovery amount, interval, stat formula or introduction date. It therefore cannot authenticate September 2002 CBT recovery.
Its EXP examples give a first attacker 66.6 versus 33.3 for equal damage, and add 10% monster EXP per additional attacker. How those examples combine, and their rounding, remain unresolved. Party sharing is described both as equal and as level-dependent. These are useful test targets, not a complete algorithm.
The recovered original Swordman and Mage guides add passive-skill tables: every 10 seconds, 5–50 HP and 3–30 SP respectively. MaxHP/MaxSP terms and effective dates are not printed. The flat HP table conflicts with the BEEhive Beta 2 MaxHP claim; the December 2003 capture dates cannot settle which rule ran in Thai CBT. Natural HP/SP formulas remain unresolved.
An independent early-beta SP candidate
A March 1, 2002 Japanese discussion, post 8, reposting an Acolyte-thread calculation, gives:
approximately 7.54 − 0.06 × INT seconds per 1 SP while standing
Its table explicitly doubles recovery while sitting. INT 10 gives 6.94 seconds per SP; INT 90 gives 2.14. These are time-per-point values, not points recovered at a fixed eight-second tick. The post does not establish the high-INT cap, the exact server/build, or a Thai deployment date. The linked original calculation has not been recovered here. Its prose calls the growth quadratic, but the stated formula describes an inverse-linear rate; retain the formula as a candidate rather than that characterization.
The July 2003 Athena source independently uses a denominator based on 7500 − 60 × INT, with a clamp, integer arithmetic and an accumulator. That resemblance strengthens the case for investigating an early timing model. It does not establish that either constant or implementation was correct for Thai beta.
| Question | Current evidence | Still missing |
|---|---|---|
| Natural SP | Early discussion describes stat-dependent time per point; historical Athena implements a similar family | Thai observation, exact constant, timer/rounding, cap, buffs |
| Natural HP | Historical Athena has a VIT/BaseLevel timing candidate | Contemporary Thai numerical measurements or original implementation |
| Weight | Thai operator guide documents suppression at red weight | Exact boundary, date of introduction, skill-specific exceptions |
| Sitting | Japanese discussion and historical Athena both accelerate recovery | Thai multiplier, timer-reset and transition behavior |
| Passive recovery skills | Original Thai guides now supply 10-second flat tables; Beta 2 notice supplies a conflicting HP candidate | Complete scaling, effective dates and independent Thai runtime corroboration |
See the existing code comparison for exact Athena accumulator semantics and modern rAthena/Hercules formulas. This research does not justify choosing one candidate as authentic yet.
Item and card descriptions may be older than their capture
The operator’s recovery-item table, preserved December 13, 2003, lists these candidates:
| Item | Approximate recovery stated | Price column, zeny | Weight column |
|---|---|---|---|
| Red Potion | 30 HP | 50 | 10 |
| Orange Potion | 50 HP | 200 | 13 |
| Yellow Potion | 80 HP | 550 | 16 |
| White Potion | 120 HP | 1200 | 20 |
| Blue Potion | 40 SP | 5000 | 20 |
| Milk | 15 HP | 25 | 5 |
The descriptions say “approximately.” They do not supply a random healing range or its VIT/INT/skill scaling. The price column does not prove actual availability at a particular NPC. No effective patch/version is identified.
The neighboring card table gives Sohee MaxSP +25%, Eggyra SP recovery +25%, and Willow MaxSP +30. These are archived documentation candidates. Neighboring weapon descriptions still mark some features unimplemented, so a December 2003 capture must not be treated as a synchronized server snapshot. Card effects, item weights and recovery formulas need independent dating.
Newly narrowed Thai dates
| Episode | Thai content | New date evidence |
|---|---|---|
| 10.1 | Einbroch | Original 2005-09-28 notice now recovered: Sakray scheduled 2005-10-03, main 2005-10-17. Main completion still unconfirmed; explicitly already current by 2005-12-08. |
| 10.4.1 | Earlier Ninja phase | Already present by the Gunslinger announcement on 2007-02-05; earlier bound than its February 27 launch schedule. |
| 12.1 | Morroc, battlegrounds and Moscovia | Same-day report says updated 2008-09-16. |
| 12.2 | Memorial Dungeons | Same-day report says updated 2008-11-18. |
The 10.1 evidence originally comprised the Thai news index, operator microsite title and dated developer interview. Round five recovered the September 28 schedule body and manual-patch body. September 28 is publication, September 29 begins preloading, and October 17 is the announced main update; none alone proves successful completion. The February 5 Gunslinger notice supplies the earlier 10.4.1 bound.
The September 16 report explicitly places Memorial Dungeons in the later 12.2 release. The November 18 report identifies that release as Endless Tower and Sealed Shrine. Their reported timers and restrictions are retained in the ledger as rules to corroborate. These reports establish a much stronger Thai sequence than copying another region’s episode date.
For 13.1, the March 15, 2009 operator index lists Ep.13.1 ASH VACUUM on March 3, linking notice 10139. That body remains unrecovered. This is an announcement-date lead, not a launch date; the May 23 report remains an observed-availability bound.
How Episode numbering is qualified
- Contemporary Thai label: the regional number appears in contemporary evidence. This still does not authenticate every mechanic in that release.
- Mirror label: early 1.0, 1.5 and 2.0 labels survive in a later notice mirror; originals remain preferable.
- Retrospective lead: a publisher sample’s indexed contents supplies Thai chapter labels for several otherwise unnumbered stages. Its direct PDF currently returns 404; only the indexed table of contents was inspected, not the full chapters.
- Contextual era: a town, balance update or seasonal event falls between identified milestones; no separate numbered episode is established.
- Unassigned: international Alpha/Beta preludes do not receive invented Thai episode numbers.
The indexed Thai book contents group Umbala/Niflheim as 7.0–8.0 and label Ayothaya 8.0. The latter remains a numbering lead, not an authenticated contemporary subversion. The original Thai combined release notice and separate town dates take precedence over imposing a universal numbering sequence.
The endpoint remains immediately before November 9, 2010. The operator’s EP13.2.1 notice includes Renewal, so that client label is excluded even though it begins with 13.2.
Remaining work before an authentic progression can be built
The main unresolved mechanics are Thai CBT HP/SP amounts and timing; passive-skill and potion scaling; complete EXP requirements and allocation; and period-specific monster, drop and spawn tables. Einbroch 10.1 now has an original October 17 main schedule, but still needs completion evidence. Exact production dates for 10.4.1, 11.1, 11.3, 13.1 and 13.2 still need deployment records. The final 2009–2010 maintenance stream remains incomplete.
Archive detail requests encountered HTTP 429; other lookups returned 404, 503 or timeouts. These are retrieval failures, not proof that the records never existed. Round-three metadata records resolved captures, hashes and failures; mechanics evidence keeps candidate values separate from verified Thai deployment rules.
The round-five date/notice packet and beta packet preserve the newer original-notice recoveries, local HTML paths and retrieval evidence. Their recovered source status does not remove the remaining effective-date or formula qualifications.