[erwan-joly]

Audit of server packets with headers m-z (and non-letter) against a real client capture (~33 MB). For every in-scope header, up to 300 distinct captured lines were deserialized with the library's Deserializer, re-serialized with Serializer, and compared byte for byte.

Capture caveats: player ids/names in the capture are already anonymized (MyIdPlayer, Me), which breaks numeric id fields on some lines (pairy, say, spk, su, tp, twk, rest). Those failures are capture artifacts, not layout mismatches. The capture is also from an older client, so fields the class has but the capture lacks were left alone.

Fixed (non-breaking)

• scr: every captured line has 7 fields; the class had 6. Added Unknow7 (default 0). Corpus snapshot updated for the extra field.
• mslot: KeyBind.SameKey (-1) was read back as CtrlA (0) because the deserializer maps every -1 enum value to the default. It now keeps -1 when the enum defines it (only affects enums with an explicit -1 member).
• Tests: CaptureAuditMzTests round-trips captured scr and mslot lines.

Breaking candidates for 27.0.0 (not changed here)

• mlobj, parcel: the sub-packet is space-separated on the wire (mlobj 1 5 6 5 3 1 0 0 0 0), but the deserializer reads a sub-packet as one dotted token, so all fields after the first are lost. Fix needs the sub-packet fields inlined into the packet.
• pst: BuffIds is List<int> but the wire carries id.level pairs (153.87 152.87); needs a sub-packet list.
• umi: wire is vnum.time in one token (umi 5060.34); class has two space fields.
• qslot: serializer drops the . after the optional Type (index 0) of QsetClientSubPacket (1.1.2 -> 11.2); serializer-level bug, also touches the shared sub-packet.

Table

header	lines checked (total / distinct)	status
m_list	770 / 101	mismatch left: duplicate header dispatch picks MlistMaterialsPacket; MlistPacket layout matches
mall	1 / 1	ok
mapout	1293 / 2	ok (one line has a stray trailing space)
minfo	1 / 1	mismatch left: -1.0/0 tokens, single sample
mlinfo	40 / 13	mismatch left: older client sends music name text, not id
mlobj	1 / 1	mismatch left (breaking, see above)
mlobjlst	43 / 10	mismatch left: 4 lines carry anonymized/odd values in list entries
mlpt	1 / 1	ok
ms_c	242 / 2	ok
msg	471 / 212	ok
mslot	12 / 2	fixed
mv	879496 / 300	ok
mz	13 / 7	ok
n_inv	214 / 117	mismatch left: list items have two shapes (6 fields for type 0, 5 otherwise)
npc_req	998 / 300	mismatch left: older client omits trailing field
npinfo	67 / 1	ok
OK	40 / 2	ok (one line has a capture comment)
out	877 / 300	ok
p_clear	115 / 1	ok
pairy	1301 / 35	ok (3 lines anonymized ids)
parcel	50 / 37	mismatch left (breaking, see above)
pidx	4 / 4	ok
pinit	1623 / 20	mismatch left: older client omits trailing fields
post	8 / 5	mismatch left: anonymized/non-UTF8 text, needs more samples
pski	1 / 1	ok
pst	12157 / 300	mismatch left (breaking, see above)
qna	30 / 11	mismatch left: embedded client packets (#arena, #guri) not resolvable from server types
qset	50 / 20	mismatch left: 5 lines use space-separated tail, meaning unclear
qslot	894 / 45	mismatch left (serializer bug, see above)
qsti	4 / 4	ok
qstlist	44 / 16	ok
rage	27 / 10	ok
raid	1023 / 300	mismatch left: multi-shape packet (type 0 member list, type 3 16-17 fields); not modelled
raidbf	7 / 3	ok (trailing space in capture)
rank_cool	36 / 1	ok
rboss	311 / 188	ok
rbr	1 / 1	mismatch left: single sample
rc	1683 / 300	ok
rc_blist	18 / 18	mismatch left: `
rc_reg	1 / 1	ok
rc_scalc	1 / 1	mismatch left: one sample has a trailing item name; not enough samples to add a field
rc_slist	16 / 12	mismatch left: nested ^ item data, anonymized names
rdi	1 / 1	ok
req_info	644 / 287	mismatch left: older client omits trailing field
rest	362 / 241	ok (11 lines anonymized ids)
revive	163 / 86	ok
rp	3 / 1	mismatch left: embedded client packet
rsfi / rsfn / rsfp	37 / 22 / 1279	ok
s_memo	4 / 2	ok
say	1376 / 300	ok (34 lines anonymized ids)
sc	1352 / 43	ok
sc_n	248 / 100	mismatch left: empty equipment written -1 on wire vs -1.0.0; older client omits trailing fields
sc_p	747 / 300	mismatch left: older client omits trailing fields
sc_p_stc	58 / 3	mismatch left: older client omits trailing field
scp	220 / 2	ok
scr	41 / 3	fixed
script	10 / 2	mismatch left: older client omits trailing field
sd	3 / 3	ok
shop	1595 / 292	ok
shopclose	37 / 1	mismatch left: older client sends no field
ski	67 / 11	ok
sopen	1 / 1	ok
sp	661 / 300	ok
spk	161 / 159	ok (4 lines anonymized ids)
sqst	257 / 21	ok
sr	381 / 13	ok
st	21824 / 300	mismatch left: older client omits trailing fields
stat	26978 / 300	ok
su	14972 / 300	ok (failures are anonymized ids)
talk	22 / 22	mismatch left: first field empty on the wire
target / targetoff	12 / 1	ok
tc_info	5 / 5	mismatch left: older layout, needs review
tit	40 / 10	mismatch left: older client sends class name text
titinfo	2 / 2	ok
title	3 / 3	mismatch left: one line has an empty sub-field (0.)
tp	331 / 278	ok (31 lines anonymized ids)
twk	36 / 12	mismatch left: older client omits trailing fields; rest anonymized
umi	30 / 7	mismatch left (breaking, see above)
vb	498 / 15	ok (one line has a capture comment)
wopen	225 / 12	mismatch left: first field empty on the wire
wp	887 / 92	ok
zzim	36 / 1	ok

Headers with no captured line (not checked): minipet, mlinfobr, mlintro, mlo_, modal, modali, msgi, msgi2, n_inv_item, n_inv_skills, NsTeST, p_sex, pet_cool2, petski, pflag, pjoin, psd, pstash, pstash_all, qnai, qnai2, qnamli, qnamli2, qr, raidopen, rc_buy, s_memoi, s_memoi2, say_p, sayi, sayi2, sayitem, sayitemt, sayt, scene, shop_end, skyinit, slinfo, stash, stash_all, stbm, stp, stp2, stpm, success, ta_, tbf, today_stc, useobj.

Summary

85 in-scope headers had captured lines; 49 round-trip cleanly (or fail only on anonymized ids, capture comments or a stray trailing space), 2 are fixed non-breakingly (scr, mslot), and the rest are left with reasons. Five clear layout mismatches (mlobj, parcel, pst, umi, qslot) need breaking changes and are grouped above for 27.0.0 / #574. All 196 tests pass.

https://claude.ai/code/session_01CUfJ45wGkdbqEmzdtJSGhD

---

Generated by Claude Code

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 59 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: NosCoreIO/NosCore.Packets
• Branch: master
• File: .coderabbit.yaml
• Commit: f52ea98
• Layers: Repository YAML
• Review profile: CHILL
• Plan: Advanced
• Run ID: fc7df583-97d0-4e2d-85e6-b28362220e83

Commits

Reviewing files that changed from the base of the PR and between f52ea98 and ebf3f5c.

Files selected for processing: 4

• src/NosCore.Packets/Deserializer.cs
• src/NosCore.Packets/ServerPackets/Player/ScrPacket.cs
• test/NosCore.Packets.Tests/CaptureAuditMzTests.cs
• test/NosCore.Packets.Tests/SerializerCorpusTests.ExpressionTreeSerializerOutputIsPinnedForEveryPacket.verified.txt

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn