AGENTS.md

@@ -12,7 +12,7 @@ not for structure or style.
 
 ## Goals
 
-The ten goals, in priority order, live in
+The goals, in priority order, live in
 [README.md](https://gitea.larvit.se/larvit/smpp-js/src/branch/main/README.md#goals) — they say where this library is heading, which an outside
 reader judges it by. The README states the audience alongside them. Everything below cites a goal by
 number.
@@ -86,8 +86,9 @@ src/
 		types.ts         Wire types: int8/int16/int32/string/cstring/buffer/arrays
 ```
 
-Dependency direction is one way: `defs` knows nothing above it, `pdu` uses `defs`, `session` uses
-`pdu`, and `client`/`server` use `session`. Nothing reaches back up.
+Imports point one way: `defs` knows nothing above it but `result.ts`, `pdu` uses `defs`, `session`
+uses `pdu`, and `client`/`server` use `session`. The one way back up is the `Session` handed to
+`IncomingRequests` and `createSms()`, imported as a type only.
 
 **Parameter order is wire order.** The key order inside `cmds.*.params` is the order the fields are
 written to and read from the buffer. Never sort those alphabetically — the alphabetical-ordering
@@ -103,20 +104,12 @@ docker compose run --rm node npm test
 docker compose run --rm node npm run build
 ```
 
-- Tests are `.ts` and run directly under Node's type stripping — no build step in the dev loop.
 - Source imports use `.ts` extensions; `rewriteRelativeImportExtensions` emits `.js` into `dist`.
 - `erasableSyntaxOnly` is on, so no enums, no namespaces, no parameter properties. Use `as const`
   objects plus union types.
-- The published floor is Node 18, but the dev container runs Node 24 (type stripping needs it). CI
-  compiles the tests and runs them on 18, every LTS above it, and current, so the floor is
-  verified rather than asserted.
+- The published floor is Node 18; the dev container runs Node 24 because type stripping needs it.
 - `typescript` is pinned to the 6.x line because `typescript-eslint` peer-requires `<6.1.0`. Move to
   TypeScript 7 once that constraint lifts.
-- GitHub mirrors Gitea through `.gitea/workflows/mirror.yaml`, which never prunes, and
-  `mirror-delete.yaml`, one run per deleted ref. A delete run that fails or outlives Gitea's queue
-  timeout, or a push run that cloned before the delete, leaves the ref on GitHub until the delete
-  run is re-run. Accepted: a stale ref there is harmless, and refs only GitHub has must survive.
-  Maintainer's call, 2026-09-14; valid while nothing deploys from GitHub.
 
 ## Defects found in 0.4.0
 
@@ -152,13 +145,6 @@ Confirmed by reading the 0.4.0 source; each row has a regression test naming the
 | `ESME_RINVBCASTCHANIND` typo | Defined as `0x011`, three hex digits; the spec value is `0x0112` |
 | Every response carries a message id | `session.js` builds `params = {'message_id': …}` for every response it sends, `deliver_sm_resp` included; SMPP 3.4 4.6.2 makes that field unused and NULL, and Jasmin closes the connection on one |
 
-## Multipart sends
-
-`sendSms` puts every segment of a message on the wire together instead of waiting for each response
-in turn, so a long message costs one round trip rather than one per segment. Nothing on the
-receiving side forces the order either way: this library answers each inbound segment as it arrives,
-so a peer that dispatches one request at a time is never left waiting on us.
-
 ## GSM 7-bit is sent unpacked
 
 Over SMPP the ESME puts one GSM character per octet in `short_message` and the SMSC packs it into
@@ -249,7 +235,6 @@ the file.
 - `Session` is publicly constructible, which is what makes `SessionOptions` and `ReconnectOptions`
   public too.
 - `acceptsOptionalParams()` and `bindAllows()` are predicates, not chokepoints.
-- `session.sock` is a getter over `PduTransport`.
 - Both emitters re-declare their listener methods to accept a promise.
 - `PduRefusedError` is exported, and `sessionError` names it in the event's type.
 - `bitCount()`, `encodeMessage()` and `splitMessage()` keep their total signatures, because
@@ -308,6 +293,7 @@ the file.
   `false` is the one way to turn it off.
 - A stream this library cannot frame is a dead link; one PDU it cannot parse is not.
 - A deliberate shutdown drains; an unusable link and an abort do not.
+- `sendSms()` puts every segment of a message on the wire together.
 - Every segment of a concatenated message is answered as it arrives, so `sendResp()` on one is the
   application's own signal rather than the peer's answer.
 - `server()` composes the application's `onRequest` after its own bind handling, and offers it every
@@ -331,3 +317,5 @@ the file.
 - `test/` stays flat too, and a file there is named for the question it answers rather than for the
   module it covers.
 - CI tests on Linux only; `src/` keeps off what is known to break on macOS or Windows.
+- GitHub mirrors Gitea without pruning, and a ref deleted on Gitea is deleted on GitHub by a run of
+  its own.

README.md

@@ -161,7 +161,6 @@ await smpp.close();      // stop listening, then drain and close every live sess
   `sendDlr('UNDELIVERABLE')` any other state: [Server in depth](#server-in-depth).
 - A message that arrived in several segments was answered as they arrived, so `sendResp()` there
   takes no `smsId` or refusing `status`. `sms.answeredOnArrival` says which case you are in.
-- `smpp.close()` stops listening, then drains and closes every live session.
 
 ## Errors
 
@@ -262,7 +261,7 @@ All optional. Timeouts are milliseconds.
 | `tls` | `false` | A `tls.TlsOptions` object with your certificate and key. A bare `true` is refused. |
 | `idleTimeout` | `40000` | Drop a peer that has been silent this long. |
 | `maxReassembly` | `1000` | Incomplete multipart messages held per session. |
-| `maxOctets` | `67108864` | Bytes of incomplete multipart messages held per session. |
+| `maxOctets` | `67108864` | Roughly the memory incomplete multipart messages may hold per session. |
 | `reassemblyTimeout` | `300000` | How long a late segment can still join an incomplete message. |
 | `responseTimeout`, `shutdownTimeout`, `maxOutstanding`, `log`, `signal` | as for the client | |
 
@@ -701,7 +700,7 @@ one wins. They do not override the [hard rules](https://gitea.larvit.se/larvit/s
    to the application to persist. Coordinating processes any other way is declined without a fresh
    argument each time.
 10. **It builds, tests and runs the same everywhere.** Container-only toolchain, no runtime
-   dependencies, the Node 18 floor verified in CI rather than asserted, every README example executed
+   dependencies, the Node 18 floor verified in CI, every README example executed
    by the suite.
 
 ## Audience
@@ -710,8 +709,7 @@ Who depends on this library, and what they may rely on.
 
 - **The public npm audience.** Only what `src/index.ts` exports is public; everything behind it is
   reshaped freely.
-- **Node 18 and newer, ESM only, no runtime dependencies.** The floor is verified in CI rather than
-  asserted, so the library drops into a service or a container without pulling a tree behind it.
+- **Node 18 and newer, ESM only, no runtime dependencies.**
 - **Real SMSCs and ESMEs as operators actually run them**, not a reference implementation. Jasmin,
   SMPPSim, Kannel, jsmpp, Cloudhopper, python-smpplib and php-smpp are the interop targets, and what
   they do in practice outranks what the specification says they should do.

docs/decisions.md

@@ -17,9 +17,6 @@ rule and an index of the titles below.
   sending them. Only `submit_sm`, `deliver_sm` and `data_sm` are policed by bind direction — the
   three the library dispatches by it, of which it sends the first two.
 
-- **`session.sock` is a getter over `PduTransport`.** Reading it is unchanged; assigning it no longer
-  compiles, which never rewired the handlers and so never worked.
-
 - **Both emitters re-declare their listener methods to accept a promise.** Maintainer's call,
   2026-08-27: `EventEmitter` types every listener as void-returning, so the
   `session.on('sms', async sms => …)` README documents reads as a misused promise in any strict
@@ -133,8 +130,7 @@ rule and an index of the titles below.
   rule purely additive: no PDU that parsed before reads differently now. Rejected: preferring the
   TLV, which re-reads every message a peer echoes into both. Rejected: refusing a PDU carrying both,
   which discards a message that is almost certainly present twice over, where goal 3 keeps the
-  traffic. The reassembler's octet cap already counts TLV values, so a 64 KB payload is bounded like
-  any other segment.
+  traffic.
 
 - **A segment's concatenation is read from its UDH, or from the `sar_*` TLVs where it declares none,
   and each spelling groups in a reference space of its own.** Maintainer's call, 2026-09-06, from
@@ -359,8 +355,8 @@ rule and an index of the titles below.
   it. There is one budget, 140 less the UDH, and the alphabet decides only what it is counted in, so
   Latin-1 and UCS2 both take those 134 octets — 134 characters and 67 — and it is GSM 7-bit's 153
   that is the odd number rather than the other way round. `Record<EncodingName, number>` is what makes
-  a fourth alphabet state its own. Rejected: 134 for GSM 7-bit too, which is the mistake the
-  unpacked-alphabet section above exists to stop. Accepted: a Latin-1 message past the 140 characters
+  a fourth alphabet state its own. Rejected: 134 for GSM 7-bit too, which is the mistake
+  [GSM 7-bit is sent unpacked](../AGENTS.md#gsm-7-bit-is-sent-unpacked) exists to stop. Accepted: a Latin-1 message past the 140 characters
   one SMS holds now costs more segments than it did, and `smsIds` is that much longer.
 
 - **An alphabet the caller named has to carry the message, and a time the format cannot express is
@@ -588,6 +584,10 @@ rule and an index of the titles below.
   reports each session's unfinished drain through `serverError`, because its own result says nothing
   but that the listener stopped.
 
+- **`sendSms()` puts every segment of a message on the wire together.** Goal 6: a long message costs
+  one round trip rather than one per segment. Rejected: sending each segment once the last is
+  answered, which a receiver waiting for the whole message before answering would deadlock.
+
 - **Every segment of a concatenated message is answered as it arrives, so `sendResp()` on one is the
   application's own signal rather than the peer's answer.** Maintainer's call, 2026-09-06, from the
   Jasmin interoperability phase: Jasmin dispatches one `submit_sm` per connector at a time and will
@@ -789,7 +789,7 @@ rule and an index of the titles below.
   dev image has no openssl.
 
 - **`src/` stays flat until a module has to move for another reason.** Architecture review,
-  2026-09-06: the grouping the file map above already implies — `wire/` for `pdu*` and `defs`,
+  2026-09-06: the grouping the [file map](../AGENTS.md#architecture) already implies — `wire/` for `pdu*` and `defs`,
   `link/` for `link-*`, `reconnect-*`, `pdu-transport` and `send-window`, `messages/` for `sms*`,
   `dlr*`, `message*`, `reassembly` and `udh` — rewrites every import for no change to
   `dist/index.js`, the one published entry. Valid while that map is what a reader navigates by.
@@ -809,3 +809,10 @@ rule and an index of the titles below.
   That binds what `dist/` runs; the container tooling, `interop-tests/` and the `package.json` scripts
   run on Linux by goal 10. Rejected: macOS and Windows runners, on GitHub's mirror or as Gitea
   host-mode runners on a Windows VM and a Mac.
+
+- **GitHub mirrors Gitea without pruning, and a ref deleted on Gitea is deleted on GitHub by a run of
+  its own.** Maintainer's call, 2026-09-14; valid while nothing deploys from GitHub.
+  `.gitea/workflows/mirror.yaml` never prunes, and `mirror-delete.yaml` runs once per deleted ref. A
+  delete run that fails or outlives Gitea's queue timeout, or a push run that cloned before the
+  delete, leaves the ref on GitHub until the delete run is re-run. Accepted: a stale ref there is
+  harmless, and refs only GitHub has must survive.

todo.md

@@ -262,8 +262,8 @@ and is also what the panel ranked hardest — two methods, one answer.
 - [ ] **Give `IncomingRequests` a port instead of the `Session` it drives.** It holds its owner and
       calls eight members of it 18 times, including `this.session.close()` on an inbound `unbind` —
       a collaborator ending its owner's life. `OutgoingRequests` is the mirror half of the same
-      boundary and takes no session at all. AGENTS.md's "Nothing reaches back up" is false because of
-      this, and `docs/decisions.md` already states the rule under The session's life: "a collaborator
+      boundary and takes no session at all. AGENTS.md names this as the one way back up; the port
+      removes that exception, and `docs/decisions.md` already states the rule under The session's life: "a collaborator
       that has to ask does not own its decision". It is also the missing test seam — inbound routing,
       reassembly dispatch, `onRequest` ordering and bind-direction refusal have no unit test because
       the class cannot be built without a live socket. Carry the eight members as `IncomingDeps`,
@@ -399,10 +399,6 @@ and is also what the panel ranked hardest — two methods, one answer.
       the fenced blocks at test time and assert each appears verbatim in the executed source, so an
       edit to either fails the gate.
 
-- [ ] **Correct AGENTS.md's "Nothing reaches back up".** False while `IncomingRequests` holds a
-      `Session`: either the first Locality item makes it true, or the sentence names the exception
-      until it does.
-
 - [ ] **Narrow the `src/defs/*` lint exemption to the four table files.** Its stated reason — "the
       spec tables are data: their length tracks the specification, not any complexity" — is false for
       `defs/types.ts`, which is 595 lines of wire codec with 25 functions and is the file that parses