jqxz2008dlx — 金庸群侠传 (Jin Yong Heroes Gathering, deluxe)

Archive: archives/金庸群侠传2008超豪华版.rar ("Jin Yong Heroes Gathering 2008, super deluxe edition"). Extracted mudlib root: raw/jy/ (a single top-level jy/ directory inside the archive, same layout as archive #91). Port: 40085. Status: done (boots clean, full registration flow verified 3 times with real Chinese names, post-login look/score/quit verified working each time).

What this is / lineage (confirmed via md5sum BEFORE doing anything else)

config.cfg's name field (GB18030-decoded: bd f0 d3 b9 c8 ba cf c0 b4 ab -> 金庸群侠传) matches the archive title exactly.

Confirmed byte-identical core engine to jqxz2008 (archive #91, libs/jqxz2008/), checked via md5sum on the raw archives before any conversion, per the task brief's lineage-check instruction:

| file | md5 vs #91 | md5 vs xiakexing3(#44) | |---|---|---| | adm/obj/master.c | identical | identical (matches #91's own prior finding) | | adm/single/master.c | identical | identical | | adm/daemons/logind.c | identical | identical | | adm/daemons/securityd.c | identical | differs (matches #91's prior finding that only securityd/config differ vs xiakexing3) | | adm/simul_efun/chinese.c | identical | identical | | adm/daemons/chinesed.c | identical | identical |

So this archive is confirmed to be the same underlying engine core as #91, exactly as the task brief predicted (title-family: xiakexing3(#44), #91, #92, #93, #94 all share the 金庸群侠传/侠客行 lineage). Per the task's explicit instruction, all of #91's proven fixes to these byte-identical files were ported directly rather than rediscovered from scratch (see below) — verified line-by-line via diff against #91's already-fixed work/ copies before applying each one, not assumed.

However, this is NOT a duplicate/near-duplicate of #91's overall content — it's a genuinely different content snapshot on the same engine (the same relationship as "different game version, same server software"). diff -rq of the two raw archives shows: file count differs (3773 raw files in #91 vs 3811 here), and dozens of zone/NPC/room files differ byte-for-byte between the two (combatd.c has different combat-formula tuning, chard.c/updated.c/several d/ zone files differ, several files exist only in one archive or the other e.g. clone/board/ mingjiao_b.c only here, d/mingjiao/guangmingding.c only in #91, etc). This matches the established shujian2008/sjtx2 and zhongjidiyu-family precedent: same engine, different content build.

Cross-check note for later archives (per task brief): archive #92 (金庸群侠传2008版.rar, being processed concurrently by a sibling agent into libs/jqxz2008std/ at the time of this writing) and archive #94 (金庸群侠传2015版.rar, not yet started) were NOT cross-checked against this lib (concurrent/not-yet-processed, per the task brief's own caveat) — whoever processes or reviews #94 next should md5sum its adm/obj/master.c/adm/daemons/logind.c/adm/simul_efun/ chinese.c/adm/daemons/chinesed.c against both #91's and this lib's raw copies; given the pattern held for #91 vs xiakexing3 vs this lib, there's a real chance #94 is yet another same-engine snapshot.

Small-to-medium lib: 3,811 raw files, 3,171 .lpc/.c files sweepable by lpcc_check.sh.

Fixes applied (with why) — ported from #91 after confirming applicability

For every fix below, the underlying file was diffed against #91's pre-fix/post-fix state first; only fixes confirmed applicable (identical buggy pattern actually present in this archive's copy) were applied. Nothing was applied blindly on lineage-assumption alone.

1. AGENTS.md §15h (is_chinese / check_legal_name — GBK byte-range checks silently wrong under UTF-8 strings). Confirmed via diff that adm/simul_efun/chinese.lpc and adm/daemons/logind.lpc were BYTE-FOR- BYTE identical to #91's own pre-fix originals (not just "similar") — ported #91's exact fix verbatim: - chinese.lpc's is_chinese(): GBK lead-byte range (str[0] > 160 && str[0] < 255, strlen>=2) → CJK Unified Ideographs codepoint range (str[0] >= 0x4e00 && str[0] <= 0x9fff, strlen>=1). - logind.lpc's check_legal_name(): byte-count bound strlen(name) < 2 || > 12 → character-count bound < 1 || > 6; dropped the i%2==0 && even-byte-offset gate. - Verified working: real Chinese names "秦风", "秦风二", "秦风三" all accepted on the first try across 3 separate registration runs.

2. AGENTS.md §15ae (private nomask command-hook breaks every post-login command silently): feature/command.lpc's command_hook was private nomask (same as #91); home/command.lpc's copy was already correct (nomask only, no private) — same split as #91. Dropped private from the feature/ copy. Verified: look and score both work correctly post-login in all 3 test runs.

3. AGENTS.md §15t (variant 1 + variant 3, combined): combatd.lpc DIFFERS from #91's copy in its combat-formula numbers (this is the deluxe edition's own balance tuning), but the include/inherit-ordering bug was present in BOTH: #include </quest/quest.h> (absolute path inside angle brackets, never resolved by this driver's inc_open()) placed AFTER inherit F_DBASE; in source order but the file-scope global quest_name it defines textually preceded the inherit once included — a fatal "Illegal to inherit after defining global variables". Fixed identically to #91: moved inherit F_DBASE; above all #includes, and changed #include </quest/quest.h> to the quoted absolute form #include "/quest/quest.h". Verified: score (which needs combatd) works cleanly post-fix; before the fix this would have thrown *No program in object '/adm/daemons/combatd'! exactly as it did in #91.

4. The same capitalize()-on-0 robustness bug in feature/name.lpc: byte-identical to #91's pre-fix short(int raw) fallback (capitalize (query("id")) with no guard). Applied the identical stringp(id) guard fix. Not independently re-triggered via a live crash this time (the tea-house board /clone/board/kedian_b in this archive also has no "short" set, same as #91, so this WOULD have crashed look there without the fix — pre-emptively fixed before first boot per the task's explicit "apply proactively before first boot" instruction, then confirmed look works cleanly with the fix in place).

5. d/city/npc/guidao.lpc's is_killing(who) direct-call type mismatch: byte-identical bug to #91 (is_killing(who) instead of is_killing(who->query("id"))). Fixed identically.

6. The "tang"-family NPC template's new used as a bare variable name: present identically in the same 7 files as #91 (d/quanzhou/npc/{tang, tang1,tang2,tang3}.lpc, d/taishan/{tang2,tang3,xingtang}.lpc). Renamed the local variable new → newob throughout (declaration + both usages) in all 7 files, exactly as in #91 — the actual new(...) efun calls are untouched.

7. AGENTS.md §15ac (bare SAVE_EXTENSION instead of __SAVE_EXTENSION__): found in the SAME 6 files as #91 — include/net/ftpdsupp.h, adm/daemons/network/netmail.lpc, cmds/arch/purge.lpc, d/quanzhou/npc/tang.lpc, d/quanzhou/npc/ tang3.lpc, d/taishan/tang3.lpc. Fixed identically. Left alone (same as #91): clone/npc/meng-zhu.lpc and u/jackyboy/TOKEN.C, which both #define SAVE_EXTENSION ".o" locally.

8. kungfu/class/mingjiao/mingjiao.h's string-literal-adjacent-to- macro-expansion syntax error (== HIG "明教" NOR with no + operators): present identically. Fixed to HIG + "明教" + NOR. A genuine divergence from #91 found here: this archive's raw copy of kungfu/class/mingjiao/mingjiao.h ALREADY had #include <ansi.h> at the top (confirmed via the raw, pre-conversion .c file, not just the converted copy) — #91's raw copy was missing it and needed it added. So only the syntax-error half of #91's two-part fix applied here; the <ansi.h>-add half was already unnecessary. Also found and fixed the SAME syntax bug in a SECOND, separate d/mingjiao/npc/mingjiao.h file (different file, same bug, not mentioned in #91's notes — possibly #91 didn't have this second copy hit by its lpcc sweep, or this is new deluxe-only content; either way it's the identical fix).

9. d/mingjiao/npc/changjinpeng.lpc's wrong absolute #include path (#include "/kungfu/class/mingjiao/tanzhu.h" — doesn't exist; real file is d/mingjiao/npc/tanzhu.h): present identically, fixed identically (#include "tanzhu.h").

10. d/mingjiao/npc/{fenggongying,baiguishou}.lpc's #include __DIR__"tanzhu.h": present identically in both files, fixed identically to plain quoted #include "tanzhu.h".

11. d/baituo/ouyangfeng.lpc, the same two independent pre-existing typos as #91: (a) missing semicolon after write("...") inside a switch case 0: block; (b) object me=this-player(); (subtraction typo for this_player()) in the LIVE inquiry_map() function. Fixed both. Note: this file also has a second, commented-out (/* ... */) copy of inquiry_map() containing the SAME this-player() typo at line 95 — left untouched since it's dead code inside a comment block, confirmed by reading the full comment span before deciding.

12. d/shaolin/obj/shuitan.lpc: ging_cost typo for the declared variable jing_cost: present identically at the same call site (me->receive_damage("jing", ging_cost)). Fixed identically. Note: this archive ALSO has a second, different shuitan.lpc at d/shaolin/shuitan.lpc (not present in #91) — checked it separately and confirmed it does NOT have this typo (already uses jing_cost correctly throughout) — left untouched, no fix needed there.

13. d/mingjiao/jmqshenmu.lpc genuinely truncated in the raw archive itself: confirmed via diff that this archive's raw .c file is byte-identical to #91's raw .c file (both 26 lines, ending mid- void init() { with no closing braces) — the exact same pre-existing truncation, not a conversion artifact, not something specific to this archive. Closed with an empty body exactly as in #91.

New fixes found in THIS archive (not part of #91's list — genuinely new

content/bugs specific to the deluxe edition's differing files)

14. d/baituo/obj/bowl.lpc and d/baituo/npc/obj/bowl.lpc (2 files, both named bowl.lpc in different directories, same content bug): set_name("海口大碗", "bowl") — second argument to set_name() must be an array of id strings (every sibling file in the same archive, e.g. d/baituo/npc/xiaoqing.lpc, correctly uses ({ "id1", "id2" })), but these two pass a bare string, which this driver's static type checker rejects (Bad type for argument 2 of set_name (string * vs string)) — the whole file failed to compile. Fixed to set_name("海口大碗", ({ "bowl" }) ) in both files. Neither file exists in #91's archive (deluxe-only content).

15. d/city2/npc/zhengmen.lpc genuinely truncated in the raw archive itself, in a DIFFERENT way than jmqshenmu.lpc (missing closing braces mid-function rather than mid-statement): confirmed via diff that this archive's raw .c is byte-identical to #91's raw copy of the SAME file (both 77 lines, valid_leave()'s nested if(dir== "north"){ ... } else { ... } block and the function itself are never closed) — this exact same truncation also exists, unfixed, in #91's shipped work/ copy (confirmed via diff against #91's raw and work directories — #91's own conversion never caught this one; it's not in #91's NOTES.md's fix list at all). Fixed here by adding the 3 missing closing braces (traced brace depth line-by-line with a small Python script to get the count right — first attempt under-added by one brace, caught by a second lpcc compile and corrected). This file is in a side zone (d/city2, the imperial-palace-gate NPC/room), not on the tested registration/look/score path, so it wasn't a registration blocker either way — found via the lpcc_check.sh sweep, not a live-boot crash. Cross-check flag: since #91 has this exact same latent bug still unfixed in its shipped work/ copy, this is worth a follow-up patch to libs/jqxz2008/work/d/city2/ npc/zhengmen.lpc too, though that's out of scope for this archive's own task. - After the brace fix, the file compiles but its create() still hits a runtime *Bad argument 1 to EFUN call_other() ... Got: int(0) while trying to populate its "objects" mapping (/d/city/npc/wujiang, /d/city/npc/bing — both files genuinely exist on disk) — same shape/category as the m_weapon/luguanji.lpc "call_other on 0" cases below (§6b-style isolated-compile artifact, not chased further); the syntax fix alone is the meaningful improvement (file went from "doesn't compile at all" to "compiles, loads, one populate-step artifact identical to an established benign category").

Confirmed NOT needed (byte-identical core files inherited the "not

needed" verdict from #91's own reading of the same file; independently re-verified via diff that the file text is unchanged, not just assumed)

• §4 (master's lazy security-daemon recursion): master.lpc identical to #91's already-fixed/confirmed-clean copy (diff against #91's work/adm/obj/master.lpc = empty). Same for §7, §8c, §15l, §15n/§15o, §15u, §15r, §15x, §15ao — all live in master.lpc/securityd.lpc, both confirmed byte-identical to #91's clean copies.
• §15p (DNS/intermud daemon in preload): adm/etc/preload has the same 11-entry list as #91, no network/dns_master — confirmed by direct read, not inferred from lineage.
• §15al (crypt(str, 0) footgun): logind.lpc identical to #91's already-confirmed-clean copy.
• §8e (tail() not a real efun): same single harmless hit as #91, in the same unused admin command cmds/wiz/tail.lpc.
• §15/§15b (simul_efun-based dbase architecture bug): confirmed this lib uses the same inherit F_DBASE architecture as #91/xiakexing3/ rzrmud, no global set/query/delete simul_efuns.
• Duplicate inherit F_UNIQUE; (§91's item 7): checked d/mingjiao/npc/{yinwushou,yinwulu,yinwufu}.lpc directly — only ONE inherit F_UNIQUE; in each file here, NOT duplicated in this archive (a genuine divergence from #91, where it was duplicated) — no fix needed.
• A LOOKALIKE duplicate-inherit false positive, checked and confirmed harmless: clone/misc/{roommaker,objmaker}.lpc each contain inherit ITEM;/inherit ROOM; 2-3 times when grepped, but on reading the actual file these are inside a @OBJ_CODE ... OBJ_CODE / @ROOM_CODE ... ROOM_CODE heredoc STRING LITERAL (a wizard tool that writes template .lpc files to disk for new objects/rooms) — not real compilation-unit inherits at all. Same false-positive shape as #91 (not flagged there either).
• §15z (#define nosave static shim): none found.
• Uppercase .C files: same 4 as #91, all u/jackyboy/{TOKEN, WIZ_CLOA,CLOTH1,GLASS}.C, already UTF-8, simply renamed to .lpc.

Content gaps confirmed as real archive gaps, not bugs (AGENTS.md §13)

• d/mingjiao/yuan/* (9 named "yuan" NPCs — 郑七灭/王八衰/赵敏/周五输/ 吴六破/孙三毁/钱二败/赵一伤/李四摧) reference two genuinely-missing kungfu skill files (xuanyuan-arrow, liangyi-jian), same as #91.
• d/mingjiao/npc/monkey.lpc references an undefined race (走兽, "beast"), same as #91. New in this archive: d/mingjiao/gudi3.lpc/ d/mingjiao/gudi4.lpc (rooms, not in #91's list) ALSO fail loading their room-embedded monkey NPC (bigmonkey) for the same undefined-race reason (走兽/走畜) — same root cause, different call sites.
• d/wudang/taoyuan/{tyroad4,5,6,7}.lpc: same confirmed ORPHANED early-draft duplicates of the working top-level d/wudang/tyroad{4,5, 6,7}.lpc files as #91 (broken #include, no NPC spawns, unreferenced anywhere in the tree).
• u/liujun/quest/{leftkiller,rightkiller}.lpc reference missing skills (six-chaos-sword, stormdance) — wizard scratch quest content, same category as #91's finding (only leftkiller was in #91's archive; rightkiller is new here, same shape).
• u/editer/tie-luohan.lpc: broken/unfinished wizard scratch draft (garbled variable name snamei), same as #91.
• adm/daemons/network/* (11 files: ping_q, mudlist_a, ms, userid, pingtcp, name_server, mail_serv, netmail, telnetd, pingd, inetd): confirmed not in adm/etc/preload and not referenced by anything outside this subdirectory — dead intermud subsystem, same category as #91 (11 here vs #91's ~16 — this deluxe build simply ships fewer of these dead files). adm/daemons/network/services/mudlist_a.lpc independently has the SAME reserved-word bug as the tang-family (mapping new, old;) but it's unreachable dead code — not fixed, same judgement call as leaving tail() alone.
• home/simul_efun.lpc: same pure sweep artifact as #91 (stray backup copy in a /home sandbox, not the configured simul_efun).
• clone/board/* corrupted save data: 11 board .o files fail restore_object() with either "Illegal file format" or "Invalid utf8 string while restoring dbase" (wudang_b, xingxiu_b, tiandihui_b, towiz_b, kedian_b, bonze_b, huashan_b, gaibang_b, gaibang_r, wiz_b, taohua_b) — same class of pre-existing corrupted seed data as #91's finding (also exactly 11 boards); each board's setup() already wraps the failure gracefully, confirmed non-fatal via the live look test at kedian_b (which just starts empty, exactly as in #91).
• adm/daemons/chinesed.lpc's data/e2c_dict.o restore also fails identically to #91 (same "Illegal file format" pre-existing corrupted seed data, caught non-fatally by master.lpc's preload() wrapper).
• d/npc/m_weapon/weapon/{m_club,m_blade,m_whip,m_sword,m_staff}.lpc: same this_player()-is-null-during-isolated-compile §6b artifact as #91 (5 files, identical shape).
• New content-gap findings, not in #91's archive:

- d/city2/obj/luguanji.lpc's carry_object(__DIR__"obj/diaopi")-> wield() fails with the same "call_other on 0" shape even though diaopi.lpc itself compiles cleanly standalone (confirmed via a direct isolated lpcc run) — same §6b isolated-compile-context artifact class as the m_weapon files, just via carry_object() instead of a direct this_player() call. - d/city2/lilishi.lpc references /d/city/obj/tudao, which genuinely does not exist anywhere in this archive — a real missing-content gap (§13), not an isolated-compile artifact. - d/xingxiu/npc/btshan.lpc references d/xingxiu/npc/ke.lpc, which doesn't exist (§13 content gap); its sibling snake.lpc DOES exist. - d/shaolin/cangku.lpc references d/shaolin/obj/{jitui,jiudai}.lpc, neither of which exists (§13 content gap). - d/shaolin/obj/{houdian,bailongdong}.lpc reference NPCs via __DIR__"npc/..." (i.e. d/shaolin/obj/npc/...), but the actual NPC files live one level up at d/shaolin/npc/{seng-bing3,xiang-ke}.lpc — a genuine relative-path mismatch (the NPCs exist, just not where these two rooms look for them). Left as a documented content/path bug rather than fixed, since the "right" fix (change __DIR__ to __DIR__+".." or similar) needs a judgement call about original intent and this is a single side-zone, not on the registration path. - d/wudang/fuzhen5.lpc references /u/fcxy/songqing.lpc, which doesn't exist (§13 content gap; there's even a commented-out alternate reference in the same line via CLASS_D("wudang"), suggesting the original author knew this NPC's location was in flux). - d/mingjiao/npc/yinsusu.lpc, d/taohua/npc/shoumu.lpc, d/xingxiu/npc/btshan.lpc (room-self, separately from its ke.lpc NPC-population issue above) each hit a "call_other on 0" during their own create() — not individually root-caused further (side-zone NPCs, not on the registration/look/score path); documented as the same general content-gap/§6b-adjacent category. - d/taohua/taohua_b.lpc (a bulletin board) tries to move() itself into /tmp/jungu/dating, which doesn't exist — a genuine missing room/zone reference (§13). - d/huashan/map.lpc: NOT real LPC code at all — it's a plain-text ASCII map (confirmed via file: "Unicode text... with CRLF, LF line terminators", and its content is garbled box-drawing/tile characters, not a program) that happens to carry a .c→.lpc extension from the original archive; confirmed via grep that nothing in the tree ever references d/huashan/map as an object path — dead/inert data file, not a bug, not fixed. - doc/build/devil.lpc, doc/efuns/shiwu.lpc: unreferenced anywhere (grepped), genuinely-broken example/tutorial scripts under doc/ (missing semicolons, a bare do_close where a string "do_close" was clearly intended for call_out) — never loaded in real gameplay, left as-is per the same judgement as other wizard-scratch content. - cmds/usr/to.lpc / cmds/wiz/to.lpc (2 files, identical bug): a real driver-compat compile error, NOT a content gap — me->edit( (: this_object(), ({ "do_to", me, arg }) :) ) fails with "Illegal to use local variable in functional" (this driver forbids capturing a local variable inside an unbound (: ... :) closure literal the way old-style MudOS code here does it). Other ->edit() call sites in this same lib (bboard.lpc, mailbox.lpc) pass a plain function-name symbol and this_player()/parameter-only args and compile fine, so this is specific to to.lpc's particular this_object(), array calling convention. This is a real, understood bug — but it's a non-critical wizard/player utility command (composes a multi-line message via the line editor), not on the registration/look/score path, and fixing it correctly would require restructuring the closure (e.g. to a bound lambda) rather than a one-line patch, so per AGENTS.md's "prioritize breadth over depth" policy it's documented here as a known, real, unfixed compile bug rather than patched.

Interactive test result — full registration + post-login flow

Driver booted clean (setsid nohup ... & disown, PID 922298, launched from libs/jqxz2008dlx/ so log/debug.log resolves correctly per AGENTS.md §6). Zero fatal errors in log/debug.log across the whole session — only the 2 expected non-fatal caught corrupted- save-data errors (chinesed's e2c_dict.o, kedian_b's board data, both documented above) and ordinary compile *warnings* ("Unused local variable", "Illegal to declare nosave function").

Ran the full registration flow 3 separate times in 3 separate continuous mudclient.py connections, with 3 different real Chinese names, confirming the fix is robust and not a one-off:

1. qinfengjy (English id) → y (confirm new character) → "秦风" → password → confirm → accept gift stats (y) → email → gender (m) → enters game world at 客店 → look → score → quit. All steps succeeded; look re-rendered the room cleanly; score showed the full character sheet (title 平民, age/gender/birthdate, 膂力/悟性/ 根骨/身法 stats, 精/气 bars, 食物/饮水 bars, kill count, 潜能/实战经验); quit saved and discononected cleanly with the "丢下一件布衣" message (unequipped default item on quit, expected). 2. qinfengjyer → "秦风二" → same full flow, same success. 3. qinfengsan → "秦风三" → same full flow, same success (this was the FINAL post-all-fixes confirmation run, done after the bowl.lpc/ zhengmen.lpc lpcc-sweep fixes, to make sure nothing in the late round of fixes destabilized the core path).

Sample transcript (run 1, abbreviated to the state-changing prompts):

您的英文名字：使用 qinfengjy 这个名字将会创造一个新的人物，您确定吗(y/n)？
请输入您的高姓大名，...
您的中文名字：（accepted 秦风 first try — before §15h fix this would loop-reject forever）
请设定您的密码：
请再输入一次您的密码，以确认您没记错：
...膂力[23]， 悟性[19]， 根骨[19]， 身法[19]
您接受这一组天赋吗？
您的电子邮件地址：
您要扮演男性(m)的角色或女性(f)的角色？

目前权限：(player)
[1;36m客店[2;37;0m -
    这是一家价钱低廉的客栈，生意非常兴隆。...
    这里明显的出口是 west 和 up。
  /clone/board/kedian_b [ 没有任何留言 ]
  店小二(Xiao er)
你上次连线是从0 on Wed Dec 31 16:00:00 1969

> 店小二用脖子上的毛巾抹了抹手，说道：这位小兄弟，请进请进。
[1;36m客店[2;37;0m - (look re-rendered cleanly)
> [1m【 平  民 】[2;37;0m普通百姓 秦风(Qinfengjy)
 你是一位十四岁的男性人类，...
 膂力：[ 23]  悟性：[ 19]  根骨：[ 19]  身法：[ 19]
 <精> ■■■■...  <气> ■■■■...
 战斗攻击力 1 (+0)    战斗防御力 1 (+1)
 食物：□□□...  饮水：□□□...
 你到目前为止总共杀了 0 个人，其中有 0 个是其他玩家。
 神： 0  潜能： 99 (1%)  实战经验： 0
> 你丢下一件布衣。
因为这样东西并不值钱，所以人们并不会注意到它的存在。
当你下次连线进来时，会从这里开始。
欢迎下次再来！

Note (same test-script gotcha as #91): the English id must be fresh letters-only each retry — a digit anywhere in the id (e.g. qinfengjy2) gets rejected by the English-name-must-be-letters check and, since mudclient.py sends its whole --send list regardless of prompt state, the rejection loop consumes all the scripted follow-up sends as more id attempts. Not a bug, just something to remember when re-testing.

lpcc sweep

Ran 3 times across the fix-iteration cycle via scripts/lpcc_check.sh:

• First sweep (after all core-engine + content fixes ported from #91, before discovering the deluxe-specific bowl.lpc/zhengmen.lpc bugs): 3104/3171 pass (97.9%).
• After fixing bowl.lpc (x2) and a first zhengmen.lpc brace attempt: 3106/3171 (97.95%) — zhengmen.lpc still failed (one brace short, caught by a Python brace-depth trace and fixed on the second attempt).
• Final: 3106/3171 pass (97.95%), same numeric pass count as the second sweep since zhengmen.lpc's remaining failure mode changed from a hard syntax error to a benign runtime "call_other on 0" content-gap artifact (documented above) rather than newly passing outright.

Remaining 65 failures were triaged by category (AGENTS.md §6b), all confirmed via direct source reading (not guessed): 11 corrupted pre-existing board saves, 11 dead network/intermud daemon files (not preloaded, not referenced), ~13 genuinely missing zone/skill/race/NPC content (mingjiao/yuan x9, monkey/gudi3/gudi4 undefined-race, wudang/taoyuan x4 orphaned drafts, lilishi→tudao, btshan→ke, cangku→jitui/jiudai, fuzhen5→songqing, taohua_b→missing /tmp/jungu/dating), 3 wizard-scratch quest/tool files (u/liujun/ quest/{left,right}killer, u/editer/tie-luohan), 1 dead-code-only reserved-word bug (network/services/mudlist_a.lpc, unreachable), 1 sweep artifact (home/simul_efun.lpc), 1 known-unfixed-efun hit (cmds/wiz/tail.lpc, §8e), 2 non-content-gap real driver-compat bugs left unfixed as non-critical (cmds/usr/to.lpc/cmds/wiz/to.lpc's closure-capture error), 2 non-LPC/dead data files (d/huashan/map.lpc, plus the 2 doc/ example scripts), 6 §6b isolated-compile-context artifacts (5 m_weapon files + luguanji.lpc's carry_object()-> wield()), a path-mismatch pair (houdian/bailongdong → their own npc/ subdir that doesn't exist), and zhengmen.lpc's now-syntax-clean but still content-gap-flavored remainder. None of these affect registration, look, or score — all confirmed via direct source reading, not guessed.

Memory stayed healthy throughout all 3 sweeps (free -h showed ~11-19GB available before/after each run, well clear of pressure; no other lpcc/driver sweeps were running concurrently at sweep time).

Process notes

• Driver launched via setsid nohup ~/src/fluffos/build-debug/src/driver config.fluffos > boot_stdout.log 2>&1 & disown from within libs/jqxz2008dlx/ (required for log/debug.log to resolve correctly, per AGENTS.md §6). Booted once, kept running through all 3 registration tests and both post-fix lpcc sweeps (the later bowl.lpc/zhengmen.lpc fixes only affect side-zones never visited by the live driver session, so no reboot was needed to validate them — the lpcc_check.sh sweep independently re-reads files from disk each run). Killed by its own exact PID (922298) at the end — verified gone via ps -p and ss -tlnp no longer showing port 40085 — never used a broad pkill pattern (per the explicit standing warning about other agents' concurrent drivers sharing the same command line).
• Scratch files (boot_stdout.log, lpcc_batch_raw.log) removed from the lib's top-level directory before finishing, per instructions; kept lpcc_fail.log (matches #91's own convention of keeping this one file).
• Port 40085 chosen per the task brief's explicit instruction (40084 reserved for the concurrent archive #92 agent; TODO.md's own "next free port" counter said 40084 at the time this task started, so 40085 is the next port after that reservation).

2026-07-23 re-verification pass (driver rebuild + formatter + WASM)

• LPC reformat: ran format-corpus.mjs over all of work/ (3,171 .lpc files) — 3,151 written, 0 already-idempotent, 20 refused (self-checked round-trip failures on messy legacy syntax, expected).
• Native retest against the freshly-rebuilt driver: booted clean on port 40085, zero fatal preload errors. Full registration flow with a fresh real Chinese name ("秦风肆"/qinfengjysi, distinct from all prior test ids on this lib) through the complete wizard (id → confirm → Chinese name → password → confirm → gift accept → email → gender) into the actual game world (客店); look, score (correct stats/combat-power/HP/potential), and quit (correct item-drop + clean save) all verified with real output, zero runtime errors in the session. No regressions from the reformat or the new driver binary.
• WASM test (scripts/wasm_client.js against build-wasm/src): boots cleanly, same preload warnings as native. Full registration + login + look + quit all work end-to-end under WASM, same as #91/ #92 — this codebase doesn't gate login on query_ip_number()'s format, so the documented WASM IP-check limitation doesn't apply. Same harmless cosmetic artifact as the other two variants: a fresh registration's "上次连线" timestamp shows the Unix epoch ("Wed Dec 31 16:00:00 1969") under WASM's synthetic clock — not a functional issue.

WASM-enablement pass (2026-07: loopback-allow + admin seeding)

Standard pass per AGENTS.md §1.3b/e + §1.5 (applied identically across the three jqxz2008 builds — engine files are byte-identical in this group):

• adm/daemons/band.lpc:39 is_banned(): short-circuit return 0 for loopback (127.0.0.1/127.*), empty, or non-string site values before the regexp ban-list scan. logind.lpc:67's BAN_D->is_banned(query_ip_name(ob)) gate in logon() is thereby loopback-proof (and robust against the WASM query_ip_number()/ query_ip_name() garbage-return bug).
• No uptime() startup-grace gate and no per-IP anti-flood/registration throttle exist in this lineage (checked logind.lpc's full input_to chain; the only other connection gate is the in-memory mad_lock admin lockdown flag, default off — left alone).

Admin account: fluffos / Mud@2026 / 浮浮, registered through the real native flow (id → y → Chinese name → password ×2 → gift y → email → gender m). Granted (admin) via adm/etc/wizlist (file was shipped empty; now contains fluffos (admin)), which securityd.lpc reads at create(). Verified after restart: login shows 目前权限：(admin), update /cmds/usr/score prints 重新编译成功. Save files (must be committed): work/data/user/f/fluffos.o, work/data/login/f/fluffos.o.

Retest: fresh registration (秦风/ceshizhe) end-to-end + look/score/quit clean; fluffos login + wizard update clean; debug.log shows only the known pre-existing corrupted-save restore errors (chinesed e2c dict / kedian board) documented above — zero new errors; test character saves removed.

Dual-mode verification pass (2026-07-24)

• Native (port 40085): fresh registration ceshier/秦风 end-to-end into 客店 + look/score/quit all correct; fluffos/Mud@2026 login shows (admin), update /cmds/usr/score prints 成功. debug.log: only the two documented pre-existing corrupted-save restore errors (caught by the lib's handler) — zero new errors. Driver killed by exact PID.
• WASM (build-wasm with query_ip_number/resolve fixes): fresh registration wasmceshi/秦风 end-to-end + look/score/quit all correct; second invocation: fluffos admin login (admin) + update 成功. Verdict: native OK + wasm OK. No fixes needed this pass. Test character saves removed; fluffos.o timestamp churn reverted.

深度功能测试（§10.7，本轮）：移植同引擎兄弟档案 jqxz2008 已确认的 4 个 bug

这份档案自己的 README 已经说明和"2008 加强版"（jqxz2008）、"2008 标 准版"（jqxz2008std）是完全相同的服务端引擎核心，只在门派 NPC、战斗 数值、部分地图内容上有独立调整。jqxz2008 自己已经做过一轮扎实的 §10.7 深度测试并记录了 4 个真实 bug——既然引擎核心相同，这轮直接逐一 核对这些 bug 是否也存在于 jqxz2008dlx（而不是重新从零排查），四个 全部确认存在且未修复：

1. adm/daemons/chinesed.lpc（§7.7：损坏存档数据把全局 mapping 清 零）。 data/e2c_dict.o 是损坏的二进制垃圾数据（不是合法的 mapping 字面量），create() 原本是 seteuid(getuid()); restore();—— restore() 抛出的运行时错误没有在 create() 内被捕获，会让 create() 在那一行直接中止，mapping dict 被清零。第一次真正用到 chinese()/to_chinese() 的地方（战斗/技能提示大量使用）就会崩溃。 修法和 jqxz2008 完全一致：catch(restore()); if (!mapp(dict)) dict = ([]);——单独加 mapp() 判断而不加 catch() 是不够的，因为不加 catch() 那一行代码根本不会被执行到。

2. adm/daemons/combatd.lpc/cmds/std/kill.lpc/cmds/skill/{bai, apprentice}.lpc（§7.11：缺失的 /log/nosave/ 目录让 write_file() 静默中止一个多步骤清理函数）。 work/log/nosave/ 目录在这份档案里 同样不存在。最严重的是 combatd.lpc 的 killer_reward()——这个函数 在每一次死亡时都会跑到，中间不加保护地 write_file("/log/nosave/ KILLRECORD", ...)，一旦目录不存在就会抛出未捕获异常，让 die() 在 那一行永久中止：角色的死亡清理（尸体、鬼魂状态、移动到死亡场景）全部 不会执行，角色变成"活着但气血精气归零"的破损状态，而且这个崩溃会在 之后的每一次 heart_beat() 都重新触发一次，变成死循环。用真实驱动 让测试角色（沈十）被"欧阳克"这个 NPC 杀死做了实测验证：你死了 → 干净地进入"鬼门关"死亡场景，NPC"白无常"正常打招呼，全程 debug.log 保持空白（十秒以上的心跳观察也没有延迟复现）。修法沿用 jqxz2008 已 验证过的 assure_file()（这份档案自己的 adm/simul_efun/file.lpc 里 已经有这个函数），在全部四处 write_file("/log/nosave/...", ...) 调 用前各加一行 assure_file(...)；顺手也给 bai.lpc/apprentice.lpc 里的 read_file() 结果加了 stringp() 判断（避免 atoi(0)）。

3. cmds/skill/bai.lpc/cmds/skill/apprentice.lpc（一处括号位置 错误，把 == 比较写进了 query() 的参数列表内）。 这两个完全相同 的档案里都有：

if (((string)me->query("family/master_id" == "feng qingyang")) || ...)

== 比较被夹在 query(...) 的参数括号内，所以传给 query() 的实际 是一个布尔/整数值（永远是 0），不是想要的属性名字符串——这个用来检 测"是否从风清扬门下叛出"的分支永远不可能触发。已改成先呼叫 query("family/master_id")，再拿它的返回值和字符串比较（和同一档案 里其它所有正确写法一致）。触发条件很窄（角色确实拜风清扬为师后又叛 出），本轮没有专门构造这个场景做实测，但代码结构本身已经明确证明这 是笔误而非设计判断。

以上四个 bug 全部对应到既有的 AGENTS.md §7.7/§7.11/§7.9 类别，不是新 分类，只是在同一引擎的另一个具体档案上再确认一次——这也进一步印证了 "共享同一底层引擎的多个档案（内容版本不同）值得直接核对已知 bug 列 表，而不是每份都从零排查"这条经验（和这个会话里 NT/nitan 家族的 §7.78 系列发现是同一个模式）。

未继续测试的部分

时间关系，除了已经验证的注册、战斗、死亡流程外，没有测试拜师 （bai/apprentice）本身、购物、以及触发风清扬叛出这个罕见分支—— bai/apprentice 的两处修复已经通过代码对照（逐字节对比两个档案确 认改动一致）验证，未做额外的实机拜师测试。

补充发现（来自对同引擎兄弟档案 jqxz2008std 的深度测试）：第 5 个 bug，§8.9 食物/饮水初始化检查错了对象

adm/daemons/logind.lpc 的 enter_world() 里：

if (!user->query("food") && !user->query("water") && ob->query("age") == 14) {

age 是角色本身（user）的属性，不是登录物件（ob）的属性—— ob->query("age") 永远是 0，这个分支从来没有真正执行过，每个新 角色的食物/饮水从创建起就永远是空的（score 显示空白状态条），静 默、无报错。已改成 user->query("age") == 14，用真实驱动重新注册 验证过（沈叁），score 正确显示食物/饮水两条状态全满。这个 bug 在 标准版（jqxz2008std）先发现，确认加强版/超豪华版/标准版三份档案 里 logind.lpc 这一行逐字节相同，一并同步修复。

WASM 修复摘要（迁移自 meta.json 的 group_note）

同一引擎，不同内容构建版本。

§7.86 跨库扫描修复（留言板 post 崩溃）

• BULLETIN_BOARD inherit + 多余 replace_program() 致命形状（AGENTS.md §7.86，post 命令崩溃）：全档案 19 处命中，已删除多余的 replace_program(...) 调用（保留 inherit），逐文件保留原有行尾格式（CRLF/LF 按文件原样）。本次为跨库 §7.86 扫描修复（触发原因：该 bug 已在 6+ 个互不相关的血统家族独立确认，属于近乎普遍的拷贝粘贴模式），仅做编译检查（驱动干净启动、端口正常监听），未做完整 §10.7 深度游玩测试。

深度功能测试（2026-08-13，round two，新驱动重测）

Re-tested against the freshly-rebuilt build-debug/src/driver（post 全库 quest_times/win_times %-operator 修复 + Warning/warning 驱动文本回退）。log_error()（adm/obj/master.lpc）已经在更早一 轮（引用注释里的"AGENTS.md #15w"）正确修复过，管理员账号 （fluffos/Mud@2026，adm/etc/wizlist 已有 fluffos (admin)） 也已经用真实注册流程创建并提交过，本轮只发现并修复了 log_file() 一处（和同引擎兄弟档案 jqxz2008 同一形状）。

发现并修复的 PROGRAMMING bug

1. log_file()（adm/simul_efun/file.lpc）完全没有 assure_file() 保护（AGENTS.md §7.11-class 的又一确认实例）：注册/登录本身只 写 log_file("USAGE", ...)（无子目录，本来就存在），不受影响， 但 nosave/CRASHES/nosave/PURGE/nosave/CALL_PLAYER 等管理指 令路径会在首次使用时未捕获抛出。已补上 assure_file(LOG_DIR + file);（含前向声明）。

Proactive checks（无需改动）

• win_times 修复确认存在且正确：d/city2/npc/refereew.lpc:176； 姊妹文件 u/wind/refereew.lpc:176/refereew2.lpc:184 同样已修。
• 未发现 message() simul_efun 包装函数——不适用 message()-missing-varargs 这一类 bug。
• 本次未观察到 jqxz2008 那份档案独有的 data/e2c_dict.o 存档损 坏问题——开机 preload 阶段和 log/debug.log 均干净，两份档案虽 然同引擎但各自数据文件的完整性状态并不共享。

实测过程

用已提交的 fluffos/Mud@2026 登录，score 确认"目前权限： (admin)"，update /adm/simul_efun/file（就是本轮改过的文件）确认 可正常重新编译。log/debug.log 时间戳全程未变化（Aug 5，早于本 次会话，且文件本身为空），确认无新增未捕获运行期错误。登录本身产 生的存档时间戳类微小 diff（data/{login,user}/f/fluffos.o 的 last_on 字段）已用 git checkout 撤销，不提交。驱动最终按精确 PID kill，ps -p 确认已退出。

（data/{login,user}/s/shenten.o 是此前会话遗留的未提交测试存档 ——Aug 5 mtime，早于本次会话，未受本轮任何操作影响，未触碰。）

§7.100 扫描修复（ROOM 基类多余 replace_program()）

#define ROOM "/inherit/room/room"：删除 826 处多余的、独立成行的 replace_program(ROOM);（保留 inherit ROOM;）。本库自带的房间 建造工具 clone/misc/roommaker.lpc（与 jqxz2008/jqxz2008std 同一血统的"寒山寺"结构）"造一间空房间"的 heredoc 模板本来就干净， 但"克隆我所在的房间"命令的字符串拼接模板把同一枚多余的 replace_program(ROOM); 烤进了每一个新克隆的房间，已同步修正。已 用 build-debug 驱动干净启动验证（0 个新增编译错误，端口正常监 听）；未做完整 §10.7 深度游玩测试。

§7.30 uninitialized-mapping accessor sweep (2026-08-20)

Corpus-wide mechanical sweep of the feature/skill.lpc shared-lineage bug (confirmed independently on xiakexing2017/jqxz2015/haiyang2 via round-four testing): 4 accessor(s) in this file returned a raw never-initialized mapping instance variable (defaults to int 0, not ([]), until first assigned), crashing any unguarded keys()/sizeof()/indexing caller for a fresh/untrained character. Fixed at the accessor level (mapp(x) ? x : ([])) per the documented remedy. Verified via lpcc --batch static compile check only (not a live boot) as part of a large mechanical sweep; not individually functionally re-tested live on this lib.

深度功能测试（§10.7，round four）：三个此前未实机测试的分支全部实测

补齐本档案自己 NOTES.md 之前标记的"未继续测试的部分"——拜师 （bai/apprentice）本身、商店购买、风清扬弟子叛出——三项全部用真 实驱动、真实指令实机验证，全程 debug.log 保持空白（无新增 error:/crash）。测试角色：shensi（沈肆，男，登记密码已略去不 记录）。

标准检查清单快速确认（应已修复，仅确认不重新排查）：

• §7.90（eval-cost）：档案顶层 config.fluffos（实际开机用的 config，不是 work/config.cfg）maximum evaluation cost 已经是 5000000，work/config.cfg 里较低的 300000 是未使用的 legacy 副本。确认无需改动。
• §7.100（ROOM 多余 replace_program()）：已在更早一轮修复 （见上方章节），本轮未重新排查。
• §7.111（standard_trace() 对 error["object"]==0 崩溃）： adm/obj/master.lpc/adm/single/master.lpc 都已经是 objectp(error["object"]) ? file_name(error["object"]) : "(driver)" 的正确写法。确认无需改动。
• §7.112（death_stage() 重入保护）：d/death/npc/ {newgargoyle,wgargoyle,bgargoyle}.lpc、d/shaolin/npc/yu-zu2.lpc 逐一读过全部 death_stage() 函数体，每一个提前 return 的分支 （目标消失、非鬼魂反被杀、最终转生）都正确地先 delete_temp("death_stage_active") 再返回，没有发现遗漏的分支。 确认无需改动。
• §7.79（裸 2 参数 addn()/addn_temp()）：全档案 work/ 树内零命中，不适用。
• §7.108（踢出重登录后指令失效）：clone/user/user.lpc:110 的 reconnect() 已经无条件在最前面 enable_commands();，修复已 存在。
• §7.30（feature/skill.lpc 未初始化 mapping accessor）：见上 一节，本轮 corpus 扫描已覆盖。
• combatd.lpc 的 bounce 除零 bug：grep -n 'bounce' adm/daemons/combatd.lpc 零命中，这份档案的 combatd.lpc 根本没 有 bounce 变量，不适用。

1. 拜师（bai/apprentice）——已实机验证，多次成功，无崩溃， 师承/头衔正确更新：测试角色先在 /d/huashan/buwei2 对入门弟子路 线的岳不群 bai yue（门槛低：shen≥1000+query_int()≥25），成功 ——score 确认"华山派第十四代弟子"，师父岳不群；随后移动到 /d/ huashan/houshan 对风清扬 bai feng（门槛高：剑法/紫霞神功/身法/ 文学等技能≥130~150、shen≥100000、query_int()≥41、随机 kar 检定），管理员用 call shensi->set(...)/set_skill(...) 赋予测试 角色对应属性/技能后（任务允许的"管理员捷径进场景，真实指令走真实 代码"手法），成功——score 确认"华山派第十三代弟子"，师父风清 扬。全程无崩溃，family_name/master_id/头衔（第N代弟子）均正确 更新，验证了本档案自己 NOTES.md 里之前只做过代码对照、从未实机触 发过的 bai.lpc/apprentice.lpc 两处修复（§7.11 assure_file() + stringp() 判断、§7.9 类的括号错位修复）在真实驱动下运行正确。

2. 商店购买——已实机验证，价格扣除+实物到手全部正确：d/city/ zahuopu.lpc（杂货铺）用 list/buy 两个指令，NPC 杨永福 （d/city/npc/yang.lpc）用 F_DEALER 混入。管理员 clone /clone/ money/coin + call coin->set_amount(500) + give coin to shensi （任务允许的"clone+give 该档案真实货币物件"标准手法，give 需要 admin 和目标角色在同一房间才能成功，一开始漏了 goto 导致"这里没 有这个活人"，补上 goto /d/city/zahuopu 后成功）给测试角色 500 文 铜板；list 确认价目表，buy budai（麻布袋，标价一两白银=100文， 注意商品的 buy 用的是 set_name() 里的别名"budai"，不是档案名 "mabudai"，用错名字会被 do_buy() 正确拒绝为"你想买什么？"而非崩 溃）购买成功，i 确认物品到手（麻布袋）且找零正确（四两白银， 500-100=400=4两，数学验证一致）。无编译错误/运行时崩溃/debug.log 报错。

（过程中发现一个值得记录、但确认是设计而非 bug 的机制：feature/ dealer.lpc 的 do_buy() 第一行就有 if ((fam = this_player()->query("family")) && fam["family_name"] == "丐帮") return notify_fail("你是个穷叫化，买什麽东西！\n");——丐帮弟子被 硬性禁止在普通商店购物，这正是任务说明里"non-beggar test character"这条要求的真正来源。测试角色因为先做风清扬叛出测试而暂 时加入了丐帮，商店测试前特意再实机 bai yue 一次把门派切回华山 派，才能通过这道设计关卡；notify_fail 走的是正常拒绝路径，没有 任何错误信号，按本 session 的 scope 纪律不作修改。）

3. 风清扬弟子叛出——已验证，真实触发，命中此前只做过代码对照的 apprentice.lpc 修复分支：读 cmds/skill/apprentice.lpc（第 55 行 if ((object)ob->query_temp("pending/recruit") == me)，本档案 自己 NOTES.md 记录的括号错位修复处）确认触发条件——目标已经对我方 pending/recruit（即目标已经用 recruit 指令点名收我为徒、只等 我回拜），且我方当前 family/family_name 与目标不同才会触发"背叛 师门"文案；若我方当前 family/master_id == "feng qingyang"，还会 额外命中一段读写 /log/nosave/FENG 计数文件的分支（本档案自己 NOTES.md 记录的另一处 §7.11 assure_file() 修复，此前也只做过代 码对照，从未实机跑过）。

实测路径：先用管理员 possess feng+recruit shensi（这一步之所 以能立即完成拜师而不是仅仅"挂起"，是因为测试角色此前有一次被风清 扬本人 attempt_apprentice() 正常拒绝的 bai feng 尝试——即使被 拒绝，apprentice.lpc 也已经在拒绝之前无条件地把 pending/apprentice 挂在了测试角色身上，recruit 指令检测到这个 挂起状态会直接完成拜师，不需要再走 attempt_apprentice() 的门槛 检查），确认测试角色 family/master_id 变为 "feng qingyang" （score 显示"华山派第十三代弟子"，师父风清扬）。随后管理员移动测 试角色到 /d/gaibang/inhole，possess zuo（左全，丐帮），执行 recruit shensi——这次确认真正挂起（左全对测试角色还没有 pending/apprentice）——输出"你想要收沈肆为弟子"，正确进入 recruit.lpc 的"挂起"分支而非"立即完成"分支。退出附身后，测试角色 真实执行 bai zuo：命中目标分支，输出"风清扬现在共有0个徒弟。" （确认 /log/nosave/FENG 被 assure_file() 正确建立且未崩溃，文 件确认内容为"0"）、"你决定背叛师门，改投入左全门下！！"、"恭喜您成 为丐帮的第二十代弟子。"——family_name 从"华山派"变为"丐帮"， master_id 从风清扬变为左全，score 确认师承信息正确更新。全程 debug.log 保持空白。这是本档案自己 NOTES.md 明确记录"逐字节代码 对照验证、未做实机测试"的两处修复（括号错位分支本身 + 嵌套的 FENG 计数器写入）第一次真正被驱动执行到，确认运行时行为完全符合预期， 没有发现新 bug。

测试方法论备注（供后续同引擎档案参考）：recruit 指令的 "挂起 pending/recruit" 分支只有在目标此前没有已经对我方挂起 pending/apprentice 时才会触发；如果顺序反过来（我方已经因为一次 被拒绝的 bai 挂起了 pending/apprentice），管理员 possess 目 标后执行 recruit 会直接完成拜师（走 recruit.lpc 自己的分支）， 不会经过 apprentice.lpc 里那处待验证的 pending/recruit 分支 ——本轮测试头几次尝试就踩到了这个顺序问题（用 call feng->recruit_apprentice(shensi) 直接注入也不可行，call 指令的 参数解析器只认字面量/带引号字符串，裸标识符 "shensi" 会被当成字符 串传入触发 call_other() couldn't find object 'shensi' 运行时报 错——这是 call 这个管理指令自身已知的语法限制，不是 mudlib bug），最终靠"先用一次会被拒绝的真实 bai（这一步真实无害地挂起 了 pending/apprentice），再用管理员 recruit 补上"的顺序绕开， 才干净地复现了目标分支。

驱动进程按精确 PID 结束（kill -TERM，确认 cwd 匹配后操作），测试 角色 shensi/管理员 fluffos 的存档增量（师承变化、FENG 计数器 从 0 到 1、fluffos.o 的正常时间戳类增量）按既有约定一并提交，未 触碰此前会话遗留、未提交的 shenten.o。