torbscript/language
Watch
1
0
Fork
You've already forked language
 0
Code Issues 139 Pull requests 2 Projects Releases 18 Packages 3 Wiki Activity Actions

os: Subnet and MacAddress (step 1 of #129, OS slice 5) #528

New issue
Closed
opened 2026-10-27 03:15:26 +00:00 by orion · 10 comments
orion commented 2026-10-27 03:15:26 +00:00
Collaborator
Copy link

Goal

Step 1 of the epic #129 (slice 5 of docs/design/OS.md section 10): the two address values that std/ip does not have yet, Subnet and MacAddress, as pure values with parsing and text form. NetworkInterface (step 2 of #129) needs both: InterfaceAddress.subnet() and hardwareAddress.

Specification

  • Design record: OS.md section 5.5 (Network) ("The address values") and the slice 5 row of section 10. docs/ROADMAP.md lists std/os "the rest of their slices" (#126 to #131).
  • What changed since OS.md was written. The design proposed a new pure std/network for the five address values. URI.md section 8a decided otherwise and it is built: the pure package is std/ip (Ipv4Address, Ipv6Address, IpAddress, SocketAddress, AddressError), and std/network is the socket package that re-exports it. Three of the five values therefore exist, with their parsers, RFC 5952 display and tests; their shape as built (octets(): List<Int>, segments(): List<Int>) stands, not the Array<UInt8, 4> of the sketch.
  • Decided by the loop, see the comment on #129: Subnet goes into std/ip (it is an IP value); MacAddress is a link-layer address and std/ip holds "nothing that is not IP" (URI.md 8a), so it goes into std/os in a module of its own, not into network.trb, so that parsing a MAC does not need the std/os/network sandbox grant (OS.md section 8, which reveals where the machine is).

Acceptance criteria

  1. Subnet in std/ip (exported from std/ip/src/lib.trb, re-exported by std/network like the other names): fields address: IpAddress and prefixLength: Int, with Show, Equals, Hash, TryFrom<String, AddressError>, static fn of(address, prefixLength): Result<Subnet, AddressError>, fn contains(address: IpAddress): Bool.
  2. A Subnet is always a network: the host bits of address are zero. tryFrom and of refuse a prefix outside 0..32 (version 4) or 0..128 (version 6), a missing /, a prefix that is empty, signed, has a leading zero or a non-digit, an address with host bits set (192.168.1.23/24) and a zone; show() is the address in its own text form, /, the prefix (10.0.0.0/8, fe80::/64), and tryFrom(show()) gives the value back.
  3. contains: first and last address of a /24, one address either side, /0 holds every address of its own version and none of the other, /32 and /128 hold exactly one address, a version 4 address is never in a version 6 subnet (an IPv4-mapped address is not matched against a version 4 subnet either).
  4. MacAddress in a new module of std/os (std/os/src/macaddress.trb or the closest name the file-naming convention of that directory allows), exported from std/os/src/lib.trb: with Show, Equals, Hash, TryFrom<String, AddressError>, octets(): List<Int> (six, most significant first), static fn of(octets: List<Int>): Result<MacAddress, AddressError>. tryFrom reads six two-digit hexadecimal groups with one consistent separator, : or - (Windows writes 3C-22-FB-0A-1B-9E), either case; it refuses five or seven groups, one-digit or three-digit groups, mixed separators and a stray character. show() is lower case with colons (3c:22:fb:0a:1b:9e). The type reads nothing from the machine (no native, no capability).
  5. The IPv6 parser is audited against the spellings OS.md slice 5 calls its risk. std/ip/tests/addresses.test.trb gets a test for each of these that it lacks: ::, ::1, 1::, :: at the start, the middle and the end, upper-case and four-digit segments, an embedded dotted IPv4 tail, :: standing for exactly one zero segment (RFC 5952 section 4.2.2: shown as 0, not ::), two runs of equal length (the first collapses), 1:0:0:2:0:0:0:3, and the refusals (two ::, nine segments, a five-digit segment, a zone).
  6. Tests: Subnet in std/ip/tests/ (criteria 2 and 3), MacAddress in std/os/tests/ (criterion 4), both run on every machine.
  7. Docs: docs/standard-library/ip.md documents Subnet, docs/standard-library/os.md documents MacAddress, docs check is clean and skills/ is regenerated. OS.md is corrected in the same PR: section 3 (module tree and the "who owns what" row: the address values are std/ip's, std/network is the socket package, MacAddress is a module of std/os) and section 5.5 (the shapes as built, the sketch's Array<UInt8, 4> replaced), and the slice 5 row says Step 1 done (date) with the step list of #129, like slice 3.
  8. sh tools/gates.sh a is green in the forge CI of the PR.

Not in this step

  • The closed list of URI.md section 9 (a string literal adapting to IpAddress, Subnet, ...) is a compiler change; OS.md only calls these types "candidates". It is not added here; if the owner wants it, it is its own issue.
  • NetworkInterface, InterfaceAddress and the natives: steps 2 and 3 of #129.

Seed impact

No. Two additive types in packages the compiler does not look up by string; no syntax change, no native, no driver flag.

Size

One PR, two Conventional Commits:

  1. feat(ip): Subnet and an audit of the IPv6 parser tests (criteria 1 to 3, 5, the ip.md part of 7)
  2. feat(os): MacAddress and the corrected address records (criterion 4, the os.md part of 7, the OS.md corrections)

Pure TorbScript and docs, no runtime C.

## Goal Step 1 of the epic #129 (slice 5 of `docs/design/OS.md` section 10): the two address values that `std/ip` does not have yet, `Subnet` and `MacAddress`, as pure values with parsing and text form. `NetworkInterface` (step 2 of #129) needs both: `InterfaceAddress.subnet()` and `hardwareAddress`. ## Specification - Design record: [OS.md section 5.5 (Network)](https://git.torb.dev/torbscript/language/src/branch/main/docs/design/OS.md#55-network) ("The address values") and the slice 5 row of section 10. [docs/ROADMAP.md](https://git.torb.dev/torbscript/language/src/branch/main/docs/ROADMAP.md) lists `std/os` "the rest of their slices" (#126 to #131). - **What changed since OS.md was written.** The design proposed a new pure `std/network` for the five address values. URI.md section 8a decided otherwise and it is built: the pure package is `std/ip` (`Ipv4Address`, `Ipv6Address`, `IpAddress`, `SocketAddress`, `AddressError`), and `std/network` is the socket package that re-exports it. Three of the five values therefore exist, with their parsers, RFC 5952 display and tests; their shape as built (`octets(): List<Int>`, `segments(): List<Int>`) stands, not the `Array<UInt8, 4>` of the sketch. - Decided by the loop, see the comment on #129: `Subnet` goes into `std/ip` (it is an IP value); `MacAddress` is a link-layer address and `std/ip` holds "nothing that is not IP" (URI.md 8a), so it goes into `std/os` in a module of its own, not into `network.trb`, so that parsing a MAC does not need the `std/os/network` sandbox grant (OS.md section 8, which reveals where the machine is). ## Acceptance criteria 1. `Subnet` in `std/ip` (exported from `std/ip/src/lib.trb`, re-exported by `std/network` like the other names): fields `address: IpAddress` and `prefixLength: Int`, `with Show, Equals, Hash, TryFrom<String, AddressError>`, `static fn of(address, prefixLength): Result<Subnet, AddressError>`, `fn contains(address: IpAddress): Bool`. 2. A `Subnet` is always a network: the host bits of `address` are zero. `tryFrom` and `of` refuse a prefix outside `0..32` (version 4) or `0..128` (version 6), a missing `/`, a prefix that is empty, signed, has a leading zero or a non-digit, an address with host bits set (`192.168.1.23/24`) and a zone; `show()` is the address in its own text form, `/`, the prefix (`10.0.0.0/8`, `fe80::/64`), and `tryFrom(show())` gives the value back. 3. `contains`: first and last address of a `/24`, one address either side, `/0` holds every address of its own version and none of the other, `/32` and `/128` hold exactly one address, a version 4 address is never in a version 6 subnet (an IPv4-mapped address is not matched against a version 4 subnet either). 4. `MacAddress` in a new module of `std/os` (`std/os/src/macaddress.trb` or the closest name the file-naming convention of that directory allows), exported from `std/os/src/lib.trb`: `with Show, Equals, Hash, TryFrom<String, AddressError>`, `octets(): List<Int>` (six, most significant first), `static fn of(octets: List<Int>): Result<MacAddress, AddressError>`. `tryFrom` reads six two-digit hexadecimal groups with one consistent separator, `:` or `-` (Windows writes `3C-22-FB-0A-1B-9E`), either case; it refuses five or seven groups, one-digit or three-digit groups, mixed separators and a stray character. `show()` is lower case with colons (`3c:22:fb:0a:1b:9e`). The type reads nothing from the machine (no native, no capability). 5. The IPv6 parser is audited against the spellings OS.md slice 5 calls its risk. `std/ip/tests/addresses.test.trb` gets a test for each of these that it lacks: `::`, `::1`, `1::`, `::` at the start, the middle and the end, upper-case and four-digit segments, an embedded dotted IPv4 tail, `::` standing for exactly one zero segment (RFC 5952 section 4.2.2: shown as `0`, not `::`), two runs of equal length (the first collapses), `1:0:0:2:0:0:0:3`, and the refusals (two `::`, nine segments, a five-digit segment, a zone). 6. Tests: `Subnet` in `std/ip/tests/` (criteria 2 and 3), `MacAddress` in `std/os/tests/` (criterion 4), both run on every machine. 7. Docs: `docs/standard-library/ip.md` documents `Subnet`, `docs/standard-library/os.md` documents `MacAddress`, `docs check` is clean and `skills/` is regenerated. OS.md is corrected in the same PR: section 3 (module tree and the "who owns what" row: the address values are `std/ip`'s, `std/network` is the socket package, `MacAddress` is a module of `std/os`) and section 5.5 (the shapes as built, the sketch's `Array<UInt8, 4>` replaced), and the slice 5 row says **Step 1 done (date)** with the step list of #129, like slice 3. 8. `sh tools/gates.sh a` is green in the forge CI of the PR. ## Not in this step - The closed list of URI.md section 9 (a string literal adapting to `IpAddress`, `Subnet`, ...) is a compiler change; OS.md only calls these types "candidates". It is not added here; if the owner wants it, it is its own issue. - `NetworkInterface`, `InterfaceAddress` and the natives: steps 2 and 3 of #129. ## Seed impact No. Two additive types in packages the compiler does not look up by string; no syntax change, no native, no driver flag. ## Size One PR, two Conventional Commits: 1. `feat(ip): Subnet and an audit of the IPv6 parser tests` (criteria 1 to 3, 5, the `ip.md` part of 7) 2. `feat(os): MacAddress and the corrected address records` (criterion 4, the `os.md` part of 7, the OS.md corrections) Pure TorbScript and docs, no runtime C.
orion commented 2026-10-27 03:22:16 +00:00
Author
Collaborator
Copy link

[loop:torbscript] claim implement (continue from the last checkpoint) by session:loop-torbscript-20261027 until 2026-10-27T09:30Z

[loop:torbscript] claim implement (continue from the last checkpoint) by session:loop-torbscript-20261027 until 2026-10-27T09:30Z
orion commented 2026-10-27 03:24:45 +00:00
Author
Collaborator
Copy link

[loop:torbscript] implement: checkpoint, next: continue implement (worker 1 ended at context limit; roughly 100k tokens used)

Branch loop/528-subnet-mac (pushed, no PR yet), commit 8e909e5f. The container has no torb binary, so nothing is compiled or formatted yet: everything below is verified by CI only.

Done

  • Subnet in std/ip/src/address.trb, exported from std/ip/src/lib.trb and re-exported by std/network/src/lib.trb.
  • Decided by the loop: Subnet is a capsule (data-or-capsule.md: private addressValue and prefixLengthValue, accessors address() and prefixLength(), ways in Subnet.of and tryFrom), not the public fields of the OS.md sketch. With public fields Subnet(address, 99) would build a value that is not a network and breaks criterion 2 ("always a network") and contains. Step 2 of #129 only needs subnet(), address() and prefixLength(). The deviation goes into the PR body and OS.md section 5.5.
  • contains and the host-bit check work on groups (octets of 8 bits, segments of 16 bits) with small shifts only, so there is no 64-bit mask overflow.

Next (in this order)

  1. std/ip/tests/addresses.test.trb: Subnet tests (criteria 2 and 3) and the IPv6 audit tests of criterion 5 (the existing file already covers ::1 ... partly; add what is missing: ::, 1::, :: at start, middle and end, upper case, four-digit segments, dotted tail, :: for exactly one zero segment shown as 0, two equal runs, refusals).
  2. std/os/src/macaddress.trb (capsule private value: Int with 48 bits, octets(), of(List<Int>) refusing a count other than six and an octet outside 0..255, tryFrom with one consistent : or -, two hex digits per group, show() lower case with colons; use AddressError from "std/ip", copy the small hexadecimalDigit helper), public use MacAddress from "./macaddress" in std/os/src/lib.trb, tests in std/os/tests/macaddress.test.trb.
  3. Docs: docs/standard-library/ip.md (Subnet in keywords, import line, torb:declaration block and prose; the declaration block must match torb docs declarations --check) and os.md (MacAddress); OS.md sections 3, 5.5 and the slice 5 row ("Step 1 done (2026-10-27)"); skills/torbscript-networking/references/standard-library/ip.md and the matching os page under skills/torbscript-standard-library regenerated (no local torb: edit the copies like the docs and let docs skill --check in CI show any diff).
  4. Two Conventional Commits as in the issue (squash the wip commit), torb format --check layout by hand (120 columns, command calls), push, open the PR Closes #528, ci-wait.sh, fix reds.
[loop:torbscript] implement: checkpoint, next: continue implement (worker 1 ended at context limit; roughly 100k tokens used) Branch `loop/528-subnet-mac` (pushed, no PR yet), commit 8e909e5f. The container has no `torb` binary, so nothing is compiled or formatted yet: everything below is verified by CI only. **Done** - `Subnet` in `std/ip/src/address.trb`, exported from `std/ip/src/lib.trb` and re-exported by `std/network/src/lib.trb`. - Decided by the loop: `Subnet` is a capsule (`data-or-capsule.md`: private `addressValue` and `prefixLengthValue`, accessors `address()` and `prefixLength()`, ways in `Subnet.of` and `tryFrom`), not the public fields of the OS.md sketch. With public fields `Subnet(address, 99)` would build a value that is not a network and breaks criterion 2 ("always a network") and `contains`. Step 2 of #129 only needs `subnet()`, `address()` and `prefixLength()`. The deviation goes into the PR body and OS.md section 5.5. - `contains` and the host-bit check work on groups (octets of 8 bits, segments of 16 bits) with small shifts only, so there is no 64-bit mask overflow. **Next (in this order)** 1. `std/ip/tests/addresses.test.trb`: Subnet tests (criteria 2 and 3) and the IPv6 audit tests of criterion 5 (the existing file already covers `::1` ... partly; add what is missing: `::`, `1::`, `::` at start, middle and end, upper case, four-digit segments, dotted tail, `::` for exactly one zero segment shown as `0`, two equal runs, refusals). 2. `std/os/src/macaddress.trb` (capsule `private value: Int` with 48 bits, `octets()`, `of(List<Int>)` refusing a count other than six and an octet outside 0..255, `tryFrom` with one consistent `:` or `-`, two hex digits per group, `show()` lower case with colons; `use AddressError from "std/ip"`, copy the small `hexadecimalDigit` helper), `public use MacAddress from "./macaddress"` in `std/os/src/lib.trb`, tests in `std/os/tests/macaddress.test.trb`. 3. Docs: `docs/standard-library/ip.md` (Subnet in keywords, import line, `torb:declaration` block and prose; the declaration block must match `torb docs declarations --check`) and `os.md` (MacAddress); OS.md sections 3, 5.5 and the slice 5 row ("Step 1 done (2026-10-27)"); `skills/torbscript-networking/references/standard-library/ip.md` and the matching os page under `skills/torbscript-standard-library` regenerated (no local `torb`: edit the copies like the docs and let `docs skill --check` in CI show any diff). 4. Two Conventional Commits as in the issue (squash the wip commit), `torb format --check` layout by hand (120 columns, command calls), push, open the PR `Closes #528`, `ci-wait.sh`, fix reds.
orion commented 2026-10-27 03:28:07 +00:00
Author
Collaborator
Copy link

[loop:torbscript] release implement: the continuation worker after the checkpoint above died on an account usage limit before its first step (no new commit; branch loop/528-subnet-mac at 8e909e5f unchanged). The next tick respawns on the next usable profile from the checkpoint above. next: implement (continue)

[loop:torbscript] release implement: the continuation worker after the checkpoint above died on an account usage limit before its first step (no new commit; branch loop/528-subnet-mac at 8e909e5f unchanged). The next tick respawns on the next usable profile from the checkpoint above. next: implement (continue)
orion commented 2026-10-27 03:28:25 +00:00
Author
Collaborator
Copy link

[loop:torbscript] claim implement (continuation 1, respawn pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027 until 2026-10-27T09:45Z

[loop:torbscript] claim implement (continuation 1, respawn pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027 until 2026-10-27T09:45Z
orion commented 2026-10-27 03:32:16 +00:00
Author
Collaborator
Copy link

[loop:torbscript] implement: checkpoint (continuation 1), next: continue implement (context guard)

Branch loop/528-subnet-mac pushed (HEAD is the commit "wip(os): shorten a doc line", main has not moved, no rebase needed), worktree code/wt/torbscript-language-528-subnet-mac (ws wt new torbscript/language 528 subnet-mac --branch loop/528-subnet-mac). No PR yet. No torb binary in the container, so nothing is compiled or formatted: CI only. Push needs git -c http.extraHeader="Authorization: token $(cat ~/.secrets/torbdev-token)" push origin loop/528-subnet-mac.

Done since the last checkpoint (4 wip commits on top of 142293fd)

  • std/ip/tests/subnets.test.trb (new): reading/showing, refusals, of, contains incl. /0, /32, /128, mid-octet prefixes, mapped addresses (criteria 2, 3, 6).
  • std/ip/tests/addresses.test.trb: group "IPv6 spellings" with the audit of criterion 5 (:: positions, case and four-digit segments, dotted tail, one-zero :: shown as 0, equal runs, refusals) and a segmentsVersion6 helper.
  • std/os/src/macaddress.trb (capsule, of, octets, show, tryFrom, copied hexadecimalDigit), public use MacAddress from "./macaddress" plus a doc bullet in std/os/src/lib.trb, std/os/tests/macaddress.test.trb (criteria 4, 6).

Next, in this order

  1. Check how the capability of a std/os module is derived (std/os/entropy is "a capability of its own"; my grep for os/entropy outside docs/.trb found nothing in the compiler, so look in compiler/ for entropy capability tables and in tools/): MacAddress must not make an importer touch the os capability if the mapping is per module; the issue wants "no capability". If a table exists, add std/os/macaddress as capability-free, else say so in the docs.
  2. Docs: docs/standard-library/ip.md (Subnet: keywords, import line, torb:declaration block that matches torb docs declarations --check, prose) and os.md (MacAddress; keep the import line style of Entropy); OS.md sections 3, 5.5 (shapes as built, Subnet is a capsule with accessors, Array<UInt8, 4> sketch replaced) and the slice 5 row "Step 1 done (2026-10-27)"; hand-copy the same changes into skills/torbscript-networking/references/standard-library/ip.md and the os page under skills/torbscript-standard-library (no local torb; docs skill --check in CI shows diffs).
  3. Squash the 4 wip commits into the two Conventional Commits of the issue (1: feat(ip): Subnet and an audit of the IPv6 parser tests, 2: feat(os): MacAddress and the corrected address records), git push --force-with-lease to the loop branch, open the PR Closes #528 (deviation note: capsule instead of public fields), ci-wait.sh, fix reds (expect formatter diffs, e.g. multi-line lists, and possibly parts[8..] in IPv6 show() for a trailing zero run, covered by the new test 1:2:3:4:5:0:0:0).
[loop:torbscript] implement: checkpoint (continuation 1), next: continue implement (context guard) Branch `loop/528-subnet-mac` pushed (HEAD is the commit "wip(os): shorten a doc line", main has not moved, no rebase needed), worktree `code/wt/torbscript-language-528-subnet-mac` (`ws wt new torbscript/language 528 subnet-mac --branch loop/528-subnet-mac`). No PR yet. No `torb` binary in the container, so nothing is compiled or formatted: CI only. Push needs `git -c http.extraHeader="Authorization: token $(cat ~/.secrets/torbdev-token)" push origin loop/528-subnet-mac`. **Done since the last checkpoint** (4 wip commits on top of 142293fd) - `std/ip/tests/subnets.test.trb` (new): reading/showing, refusals, `of`, `contains` incl. /0, /32, /128, mid-octet prefixes, mapped addresses (criteria 2, 3, 6). - `std/ip/tests/addresses.test.trb`: group "IPv6 spellings" with the audit of criterion 5 (`::` positions, case and four-digit segments, dotted tail, one-zero `::` shown as `0`, equal runs, refusals) and a `segmentsVersion6` helper. - `std/os/src/macaddress.trb` (capsule, `of`, `octets`, `show`, `tryFrom`, copied `hexadecimalDigit`), `public use MacAddress from "./macaddress"` plus a doc bullet in `std/os/src/lib.trb`, `std/os/tests/macaddress.test.trb` (criteria 4, 6). **Next, in this order** 1. Check how the capability of a `std/os` module is derived (`std/os/entropy` is "a capability of its own"; my grep for `os/entropy` outside docs/.trb found nothing in the compiler, so look in `compiler/` for `entropy` capability tables and in `tools/`): `MacAddress` must not make an importer touch the `os` capability if the mapping is per module; the issue wants "no capability". If a table exists, add `std/os/macaddress` as capability-free, else say so in the docs. 2. Docs: `docs/standard-library/ip.md` (Subnet: keywords, import line, `torb:declaration` block that matches `torb docs declarations --check`, prose) and `os.md` (MacAddress; keep the import line style of `Entropy`); OS.md sections 3, 5.5 (shapes as built, Subnet is a capsule with accessors, `Array<UInt8, 4>` sketch replaced) and the slice 5 row "Step 1 done (2026-10-27)"; hand-copy the same changes into `skills/torbscript-networking/references/standard-library/ip.md` and the os page under `skills/torbscript-standard-library` (no local `torb`; `docs skill --check` in CI shows diffs). 3. Squash the 4 wip commits into the two Conventional Commits of the issue (1: `feat(ip): Subnet and an audit of the IPv6 parser tests`, 2: `feat(os): MacAddress and the corrected address records`), `git push --force-with-lease` to the loop branch, open the PR `Closes #528` (deviation note: capsule instead of public fields), `ci-wait.sh`, fix reds (expect formatter diffs, e.g. multi-line lists, and possibly `parts[8..]` in IPv6 `show()` for a trailing zero run, covered by the new test `1:2:3:4:5:0:0:0`).
orion commented 2026-10-27 03:34:27 +00:00
Author
Collaborator
Copy link

[loop:torbscript] claim implement (continuation 2 of 3, pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027 until 2026-10-27T09:45Z

[loop:torbscript] claim implement (continuation 2 of 3, pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027 until 2026-10-27T09:45Z
orion commented 2026-10-27 03:39:14 +00:00
Author
Collaborator
Copy link

[loop:torbscript] implement: checkpoint (continuation 2), next: continue implement (context guard; PR open, forge CI running)

PR #530 (Closes #528), branch loop/528-subnet-mac, two Conventional Commits (0147b733 feat(ip), 3a03f5ec feat(os)), rebased on main (no drift), worktree code/wt/torbscript-language-528-subnet-mac. No torb in the container: nothing compiled or formatted, the forge CI is the first run.

Done in this continuation

  • Capability: std/os/macaddress would have reached "the operating system" through capabilityOf("std/os"); added the arm ("std/os", "MacAddress") => None in compiler/src/package/archive.trb (capabilityOfImport) and a test in compiler/tests/capability.test.trb.
  • Docs: ip.md (Subnet: keywords, import, declaration block, prose, checked example), os.md (MacAddress), OS.md sections 3 and 5.5 and the slice 5 row ("Step 1 done (2026-10-27)"), summaries in the index files and the skill copies (skills/ edited by hand with the link rewrites of the generator).
  • Squashed the wip commits into the two commits, force-with-lease pushed, PR opened.

Next

  1. bash /opt/openclaw/skills/dev-workflow/scripts/ci-wait.sh -C <worktree> --timeout 3300 (a waiter was started in the previous worker; it may have ended), fix every red on the same branch. Expect: torb format --check diffs (120 columns, command-call style), docs check/docs declarations --check on the Subnet and MacAddress blocks (member order or private handling), docs skill --check diffs of the hand-edited skills/ copies (regenerate by copying the CI diff), and possibly a compile error in the unverified .trb (subnets.test.trb, macaddress.trb, IPv6 show() with a trailing zero run).
  2. When tier A is green: result comment [loop:torbscript] implement: PR open, tier A green, next: review (merge gate argus) with token usage; release the claim.
[loop:torbscript] implement: checkpoint (continuation 2), next: continue implement (context guard; PR open, forge CI running) PR https://git.torb.dev/torbscript/language/pulls/530 (`Closes #528`), branch `loop/528-subnet-mac`, two Conventional Commits (0147b733 `feat(ip)`, 3a03f5ec `feat(os)`), rebased on main (no drift), worktree `code/wt/torbscript-language-528-subnet-mac`. No `torb` in the container: nothing compiled or formatted, the forge CI is the first run. **Done in this continuation** - Capability: `std/os/macaddress` would have reached "the operating system" through `capabilityOf("std/os")`; added the arm `("std/os", "MacAddress") => None` in `compiler/src/package/archive.trb` (`capabilityOfImport`) and a test in `compiler/tests/capability.test.trb`. - Docs: `ip.md` (Subnet: keywords, import, declaration block, prose, checked example), `os.md` (MacAddress), OS.md sections 3 and 5.5 and the slice 5 row ("Step 1 done (2026-10-27)"), summaries in the index files and the skill copies (`skills/` edited by hand with the link rewrites of the generator). - Squashed the wip commits into the two commits, force-with-lease pushed, PR opened. **Next** 1. `bash /opt/openclaw/skills/dev-workflow/scripts/ci-wait.sh -C <worktree> --timeout 3300` (a waiter was started in the previous worker; it may have ended), fix every red on the same branch. Expect: `torb format --check` diffs (120 columns, command-call style), `docs check`/`docs declarations --check` on the `Subnet` and `MacAddress` blocks (member order or `private` handling), `docs skill --check` diffs of the hand-edited `skills/` copies (regenerate by copying the CI diff), and possibly a compile error in the unverified `.trb` (`subnets.test.trb`, `macaddress.trb`, IPv6 `show()` with a trailing zero run). 2. When tier A is green: result comment `[loop:torbscript] implement: PR open, tier A green, next: review (merge gate argus)` with token usage; release the claim.
orion commented 2026-10-27 03:40:21 +00:00
Author
Collaborator
Copy link

[loop:torbscript] release implement (continuation 2 of 3 collected), next: CI on PR #530 (head 3a03f5ec, all checks pending at collect time). Green tier A: hand over to review. Red: one fix worker (continuation 3, last allowed) briefed from issuecomment-7330. No worker slot free right now (2 running), so the next tick picks this up by CI state. Worker tokens: ~97k prompt/cache.

[loop:torbscript] release implement (continuation 2 of 3 collected), next: CI on PR #530 (head 3a03f5ec, all checks pending at collect time). Green tier A: hand over to review. Red: one fix worker (continuation 3, last allowed) briefed from issuecomment-7330. No worker slot free right now (2 running), so the next tick picks this up by CI state. Worker tokens: ~97k prompt/cache.
orion commented 2026-10-27 07:24:39 +00:00
Author
Collaborator
Copy link

[loop:torbscript] claim implement (fix round, red tier A on PR #530 head 3a03f5ec, pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027-0718 until 2026-10-27T13:24Z

[loop:torbscript] claim implement (fix round, red tier A on PR #530 head 3a03f5ec, pinned on anthropic:notahumanmate) by session:loop-torbscript-20261027-0718 until 2026-10-27T13:24Z
orion commented 2026-10-27 07:56:42 +00:00
Author
Collaborator
Copy link

[loop:torbscript] release implement: the fix worker timed out while waiting for CI; PR #530 head 00052388 carries its push, forge CI is queued. next: waiting (CI on PR #530); red tier A gets a further fix worker (the issue's continuation cap is spent, so the fix round continues from the PR state)

[loop:torbscript] release implement: the fix worker timed out while waiting for CI; PR #530 head 00052388 carries its push, forge CI is queued. next: waiting (CI on PR #530); red tier A gets a further fix worker (the issue's continuation cap is spent, so the fix round continues from the PR state)
argus closed this issue 2026-10-27 14:19:46 +00:00
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
loop/705-default-shadow
loop/691-tensor-core
loop/705-fix-candidate
orion/456-tierb-probe
loop/164-measure2
loop/207-measure
loop/164-prototype
probe/207-count
loop/205-measure
loop/140-hide-staged
nightly-20261103
nightly-20261031
nightly-20261029
nightly-20261027
nightly-20261026
nightly-20261025
nightly-20261024
nightly-20261023
nightly-20261021
nightly-20261020
nightly-20261015
nightly-20261014
nightly-20261013
nightly-20261011
nightly-20261010
nightly-20261009
nightly-20261008
seeds
Labels
Clear labels   
loop
Backlog item of the project loop (Orion), see CONTRIBUTING

  
ready
Refined to the definition of ready

No labels
loop
ready
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
torbscript/language#528
No description provided.