codegen: inline 1-byte tag fast path at decode call sites

Hoists wire.decode_tag's 1-byte fast path into every generated
M.X_decode while-loop, falling back to the helper for multi-byte
tags (field ids > 15). The 1-byte case covers every protobuf field
with id 1..15 and is the dominant decode dispatch in real payloads.

Header now localizes string.byte, bit.band, and bit.rshift so the
inlined ops compile to straight local calls.

Bench (Person full decode, msgs/s, median-of-3 vs proper post-h8v
3-run baseline): 10B +14.7%, 100B +8.9%, 1KB +7.5%, 10KB +7.0%,
100KB +8.5%. Full encode is flat to small (-0.1% to -2.6%) at large
sizes, plausibly from header-upvalue layout. JIT trace gate: 37/37,
all bridges still 0. Tests: 745/745.

Also documented in bench/PERF_LOG.md, including the methodology
note that h8v's earlier numbers used single-run baselines and are
therefore ~3-5% optimistic; medians-of-3 are the standard now.

beads-tarantool-protobuf-4kj

codegen: inline 1-byte varint length prefix at every LEN emit site

Eliminates the wire.encode_varint(#body) call + dispatch for every
length-delimited field in mode=full codegen. Profile flagged the
out[n] = wire.encode_varint(#_b) line as ~33% of hello.Person 1KB
encode time, with another ~17% in encode_varint dispatch — together
~50% of encode time. Lifting the 1-byte fast path (the dominant case
for proto strings and small message bodies) to the call site removes
the function frame entirely for lengths < 128.

Applied at every LEN emit site: singular/repeated message body,
singular/repeated string|bytes, packed scalar bundle, packed enum
bundle, map entry. Map value pieces (emitMapPiece message branch)
left as-is — they sit inside a single slot assignment that would
require a deeper restructure, and maps are not on the current hot
benchmark.

Results (hello.Person full encode, msgs/s): 10B +6.9%, 100B +7.9%,
1KB +25.7%, 10KB +48.1%, 100KB +31.9%. Decode flat (unchanged path).
Runtime mode flat (descriptor dispatch still calls encode_varint).
JIT trace gate: 37/37. Test suite: 745/745.

Bench history saved to bench/PERF_LOG.md with full numbers and the
workflow this iteration follows.

beads-tarantool-protobuf-h8v

bench,codec: pin proto2 paths, kill pairs() on extension hot path

JIT trace gate gains 12 proto2-specific checks (required encode/decode,
groups singular + repeated, extension encode/decode). The extension-encode
check fired NYI on bytecode 72 (ISNEXT) — pairs() over the new
extensions_by_full_name hash is the same trace-abort that gates map
encode. Fix: register_extension now also appends to extensions_list,
an array view. codec.encode_message, text.emit_message and
json.encode_message all switch to ipairs over the list. Hash tables
stay around for O(1) lookups in decode (extensions_by_id) and bracket-
name resolution (extensions_by_full_name).

Inline (full-mode) codegen learns to walk extensions too: before this
commit the inline encoder skipped t._extensions entirely (only the
runtime codec walked it), so a generated _encode silently dropped any
extension data set on the message. Add the array-walk after the field
loop and a decode_extension dispatch in the unknown-tag branch — the
inline path now matches runtime byte-for-byte.

bench/bench.lua parameterizes over a FIXTURES list so the baseline can
cover both hello.Person and a new proto2_basic.BenchPayload fixture
(required + group + extension + repeated). baseline.json restructured
under "schemas": [{schema, results}, …]; compare walks both. Fresh
numbers committed.

740/740 luatest cases pass; JIT gate 37/37; conformance 2806 binary+JSON
and 434 text-format, both 0 failures.

codegen: preserve descriptor options as `options = { ... }`

Every populated *Options message — FileOptions, MessageOptions,
FieldOptions, OneofOptions, EnumOptions, EnumValueOptions,
ServiceOptions, MethodOptions — surfaces on the generated descriptor as
a plain Lua sub-table named `options` (or `oneof_options` /
`value_options` for the per-member shapes). Standard fields use their
proto name as a bare Lua key; extensions use their fully-qualified
extension name as a bracket-quoted string key.

The walker is generic — no extension-specific code paths. Consumers
pull whatever they care about: `(google.api.http)` for REST routing,
`(versionpb.etcd_version_*)` for compatibility gates, `[deprecated =
true]` for migration tooling, and any in-house extension without pb
knowing about them. Standard fields sort alphabetically before
extensions (also alphabetical by full name) so codegen output stays
byte-identical across runs.

The `options` key is only emitted when at least one field is populated,
so option-free protos produce zero-diff output to before. Resolver
re-links *Options* messages so in-file extensions surface via the
protoreflect walker — protogen builds f.Desc before in-file extensions
are registered, and only f.Proto gets the post-pass fix-up, so we
rebuild the resolver manually.

UninterpretedOption is treated as a codegen-time error: a populated
entry means protoc couldn't resolve the extension, and emitting
opaque parser state would hide the problem.

codegen: emit strict <Type>_fields / _oneofs constants for lazy view

Lazy-view callers passing a typo'd field name to :get / :has / :set /
:clear / :which got `nil` back, indistinguishable from a legitimately-
absent optional field. Failures surfaced as missing data downstream.

Each generated message now exports a M.<Type>_fields table mapping
each field name to itself (and M.<Type>_oneofs for oneof groups),
wrapped by a new pb.field_names() helper that errors on unknown-key
reads and on any write. Routing field-name arguments through these
tables turns a typo into a load-time error at the read site.

Eager _encode / _decode keep round-tripping plain Lua tables — the
constants table is a lazy-view contract, documented in
docs/api-modes.md. README and lazy_test.lua converted to the new
pattern; three new tests cover typo / read-only / oneof-typo errors.

codegen: propagate proto comments into generated _pb.lua

Leading `//` comments on messages, enums, enum values, fields, services,
and RPC methods now surface in the generated Lua:

- Message / enum descriptions become `---` blocks above `---@class` /
  `---@alias`, so LuaLS shows them on hover.
- Per-field comments collapse onto a trailing `@ description` on the
  matching `---@field` line.
- Enum values, service banners, and per-method entries get plain `--`
  lines inside the table literals so readers without an LSP still see
  the proto-side context.

Covers both `mode=full` and `mode=runtime` (emission is mode-independent
but the new luatest group runs against both).

text: render captured unknown fields, tolerate SGROUP in skip

Two changes close the proto3 text-format conformance suite:

  1. `wire.skip_field` learns SGROUP/EGROUP. Wire 3 recurses through
     inner tags until a matching EGROUP, with field_id checked against
     the SGROUP's id. Callers (codec.lua, lazy.lua, wkt.lua, generated
     full-mode `_pb.lua`) now pass the tag's field_id so groups inside
     unknown-field skips don't error.

  2. `pb.text.encode` walks the captured `_unknown_fields` buffer when
     `opts.print_unknown_fields=true` and emits each entry in
     TextFormat numeric-field form:
       VARINT  -> "<id>: <uint64>"
       I64/I32 -> "<id>: 0x<hex>"
       LEN     -> speculative "<id> { <recurse> }"; rolls back to
                  byte-string form if the inner bytes don't parse as a
                  sub-message
       SGROUP  -> "<id> { <recurse> }" through matching EGROUP

`cmd/conformance/core.lua` threads `req.print_unknown_fields` into
`pb.text.encode` so `_Drop` tests drop unknowns and `_Print` tests
render them.

Conformance: text-format suite goes from 2 ✓ / 6 expected fails to
8 ✓ / 0 expected fails. All eight regression tests in
`conformance_test.lua` (one per upstream test, plus the fixed field-1011
tag bytes that were miscomputed earlier) now assert the target output.

codegen+wire: split length-prefix emission + 2-byte varint fast path

Two complementary encode optimizations: codegen + runtime writers no
longer pay a per-field string concat for length-prefixed fields, and
encode_varint_slow now handles 2/3/4-byte values without dropping into
the uint64 cdata path.

Composite bench/bench.lua, full mode:
  1KB encode:    229 K -> 332 K msgs/s  (1.45x)
  10KB encode:   38 K  -> 56.5 K        (1.49x)
  100KB encode:  4.1 K -> 6.7 K         (1.64x)

Runtime mode now matches full mode at 10KB+:
  1KB encode:    203 K -> 276 K  (1.36x)
  10KB encode:   35 K  -> 55 K   (1.56x)

bench/shapes_bench.lua biggest swings (both modes):
  packed-int32x1000 enc:  1.7 K  -> 13 K  (7.0-7.6x)
  packed-int32x100  enc:  21 K   -> 113 K (5.3x)
  scalar-heavy      enc:  582 K  -> 1.0 M (1.7x)

Per-helper, bench/wire_bench.lua:
  encode_varint(150)  [2-byte]:   529 ns -> 68 ns  (7.8x)
  encode_varint(20K)  [3-byte]:   841 ns -> 88 ns  (9.6x)
  encode_tag(16, LEN) [2-byte]:   531 ns -> 78 ns  (6.8x)
  encode_string(200B) [2-byte L]: 560 ns -> 109 ns (5.1x)
  encode_varint(127)  [1-byte]:    29 ns -> 28 ns  (no regression)

1. Split length-prefix emission. Length-delimited fields (string/bytes
   scalars, nested messages, packed scalars/enums) used to emit
   `tag → encode_len(body)` where `encode_len(body)` returns
   `encode_varint(#body) .. body`. The string concat allocated a copy
   of the body per field. Now the codegen and runtime writers emit
   three separate `out` slots — `tag`, `varint(#body)`, `body` — and
   let `table.concat(out)` join them in one pass at the end of encode.

   Applied to inline.go (full-mode codegen) for: singular and repeated
   nested messages, singular and repeated string/bytes scalars, packed
   scalars, packed enums. Applied to codec.lua build_writer / build_
   repeated_writer for the same set. Maps still go through encode_len
   pending a separate pass.

2. encode_varint_slow Lua-number fast paths. The outer encode_varint
   stays at the tiny `1-byte check + tail call` shape that LuaJIT
   inlines into hot traces. The slow function (not inlined into hot
   traces, so its body size is unconstrained) now handles non-negative
   Lua numbers up to 2^28 directly via bit.rshift / string.char with no
   cdata allocation. Values in [2^28, 2^53) emit one byte and recurse
   on the smaller residue. Only cdata inputs, negative Lua numbers
   (sign-extended to 10-byte varint), and the rare > 2^53 case still
   take the uint64 cdata loop. Net effect: every multi-byte varint
   encode that fits in a Lua number — including the length prefix for
   any string >= 128 bytes, every tag for field IDs >= 16, and every
   negative-zigzag sint — drops from ~500 ns to ~70 ns.

497/497 luatest pass.  23/23 jit-trace gates pass (the two new
multi-byte varint gates added in the bench infra commit confirm
encode_varint_slow JIT-compiles cleanly).  Conformance suite (binary
+ text) shows 1478 expected passes, 0 unexpected failures.

json: treat null fields as absent (and Value's null as a real value)

Per the proto3 JSON spec, a null on any field means "use the field's
default" — encoded as missing — with the lone exception of
google.protobuf.Value, where JSON null is itself a Value carrying
NullValue.NULL_VALUE.

Three coupled bugs surfaced together:

1. decode_field_value used to fall through with v = box.NULL, leaving a
   useless box.NULL sitting in the result table for scalars. Now it
   returns nil for non-Value fields, PB_NULL for Value fields.

2. decode_message's repeated and map branches called `#jv` and
   `pairs(jv)` unconditionally; a JSON-null on either type crashed with
   "attempt to get length of 'void *'". Now both branches short-circuit
   when jv is box.NULL.

3. The nil-skip checks in the decode loop (`if dv ~= nil`) and in the
   codec / inline message encoders (`if v == nil then return end`)
   evaluated TRUE on box.NULL because Tarantool's cdata __eq aliases
   it to nil. Decode now uses rawequal(dv, nil); encode special-cases
   message kind by also accepting cdata, so the Value field's
   box.NULL sentinel survives all the way through to value_encode.

Drops 3 entries from test/conformance/known_failures.txt
(AllFieldAcceptNull, WrapperTypesWithNullValue, ValueAcceptNull).
Adds 5 regression tests covering scalar / repeated / map / wrapper
null treatment and the Value-NULL_VALUE exception.

codec: recursively merge repeated singular-message wire entries

Per proto3 spec, when the same singular message field (including a oneof
branch) appears twice on the wire, the two values must merge: scalar
fields last-wins, repeated fields concatenate, sub-messages merge
recursively, maps last-wins per key. The previous reader replaced the
prev value wholesale for oneof branches and overwrote repeated/nested
fields with `prev[k] = v` even outside oneofs, losing data unique to
the first occurrence.

Adds wire.codec.merge_message(desc, prev, decoded), a descriptor-driven
recursive merge, and routes both codec.lua's reader and the inline-mode
codegen through it. WKT message fields (custom decode) keep the replace
behavior because their decoded value is not a generic Lua table.

Exposes pb.codec to the generated inline code so the helper is reachable
without a per-call require.

Drops 3 entries from test/conformance/known_failures.txt:
ValidDataOneof.MESSAGE.Merge, ValidDataOneofBinary.MESSAGE.Merge,
RepeatedScalarMessageMerge. Adds 5 regression tests covering scalar
last-wins, oneof merge, recursive sub-message merge, repeated-in-
submessage concat, and oneof sibling clearing after merge.

wire: truncate varints to 32 bits in int32/uint32/sint32/enum decode

decode_int32/uint32/sint32 returned full uint64 values when the wire input
carried bits above bit 31, corrupting re-encode on 51 conformance tests that
exercise overlong or over-range varints. Per the proto3 spec the decoder
must keep only the low 32 bits (and sign-extend for signed types).

Adds wire.varint_to_int32 / varint_to_uint32 and routes every enum-varint
decode site through them: wire.lua typed decoders, codec.lua (5 enum
sites), lazy.lua (3 sites), and the generated code via inline.go (3 sites).

Drops 51 entries from test/conformance/known_failures.txt.

codegen: text-format printer (pb.text.encode + <Msg>_text wrappers)

Adds a descriptor-driven text-format encoder mirroring `protoc --decode`
output: one field per line, 2-space indent, octal byte escapes,
`nan`/`inf` floats, `opts.single_line=true` for compact one-liners.
WKT-aware — Timestamp/Duration accept datetime cdata or {seconds,nanos},
wrappers print their unwrapped scalar, Struct/Value/ListValue walk the
tagged-table form, FieldMask flattens to `paths:` lines, Any stays
opaque.

Both codegen modes emit `M.<Type>_text(t, opts)`, surfaced as `pb.text`
on the public table. Encode-only; matching parser deferred.

Tests cover scalars, repeated, maps, oneof, optional presence, all WKT
types, single-line mode, and the generated wrapper across both modes
(74 cases). 403/403 luatest green, 19/19 jit-trace gate green.

codegen: EmmyLua type annotations for messages, enums, wrappers

Generated Lua modules now carry lua-language-server type annotations:
  ---@alias <full.Enum> integer            per enum
  ---@class <full.Message>                  per message
  ---@field <name> <type>                   per field
  ---@param / ---@return                    per wrapper

Mappings:
  bool                       -> boolean
  string / bytes             -> string
  float / double             -> number
  all int kinds              -> integer  (64-bit cdata typed as integer;
                                          LSP has no cdata model)
  enum / message             -> <full.Name>  (resolves to declared alias/class)
  repeated T                 -> T[]
  map<K,V>                   -> table<K, V>

Presence markers (trailing `?` on field name):
  proto3 explicit optional
  oneof branches             (only one is set at a time)

Wrapper signatures cover _new / _encode / _decode / _decode_lazy plus
_has_<field> / _clear_<field> on optional fields. _decode_lazy returns
pb.MessageView, which is declared inline in runtime/pb/lazy.lua along
with pb.ArrayView and pb.MapView so cross-file references resolve in
any project that requires('pb.lazy').

Class identifiers use proto full names verbatim (e.g. `hello.Person`)
so cross-file imports and WKT references both resolve to a single
declared `---@class` block — no per-module renaming needed.

Pure comment addition: 300/300 luatest + 19/19 jit-trace gate stay
green. Generated examples regenerated and committed.

lazy: zero-copy decode view (decode_lazy) with passthrough re-encode

Adds pb.decode_lazy(desc, bytes) returning a MessageView/ArrayView/MapView
that indexes the wire bytes in a single pass and decodes individual
fields only on :get / :at access. Nested messages return more lazy
sub-views; WKT descriptors (those carrying desc.decode) are
eager-wrapped so the API stays uniform.

Surface (see runtime/pb/lazy.lua):
  - MessageView: :get / :has / :which / :iter / :names / :set / :encode
  - ArrayView:   :len / :at / :iter / :tolist
  - MapView:     :get / :has / :keys / :iter / :totable

:encode is three-modes: WKT delegates to desc.encode on the materialized
table; untouched views return their original bytes verbatim
(passthrough); mixed views walk fields in id order, splicing clean
segments and re-emitting dirty ones. Sub-MessageView mutations
propagate to parent encode via a flat _sub_msg_views array (walked with
ipairs, so :is_dirty stays on a single JIT trace — pairs over a hash
is NYI in LuaJIT 2.1).

Codegen emits M.<Type>_decode_lazy in both modes as a one-line
delegation to pb.decode_lazy(<desc>, b); no inline expansion.
codec.encode_field is exposed so the lazy passthrough emitter can
splice fresh bytes for a single dirty field without rebuilding the
whole message.

Tests: 40 new lazy_test.lua cases parameterized over both codegen
modes; all 11 interop fixtures round-trip byte-equal through
decode_lazy(b):encode() in both modes. Total: 300/300 luatest, up
from 226.

Initial commit: protoc-gen-tarantool plugin + pb runtime

A protoc plugin (Go) and a pure-Lua + LuaJIT-FFI runtime that give
Tarantool a complete proto3 + gRPC stack. Two codegen modes (full
inline / runtime descriptor), 226-test luatest suite, 18-fixture
mainline-protoc interop corpus, JSON codec, well-known types,
gRPC client/server factories, runtime .proto parser, microbench
harness with allocation regression gate.

Covers PLAN.md M1-M5. Module is `pb` (not `protobuf`) to avoid
colliding with Tarantool's built-in encode-only `protobuf` module.