Browse Source

GOALS: add G-061 [proposed] pseudoconsole expect-alternative for interactive shell testing

- ConPTY (windows) + pty (unix) expect-like harness: spawn a built punkshell on a size-controlled pseudo-terminal, send keys incl control/arrow sequences, await patterns in the rendered ANSI byte stream with timeouts and forced teardown, full transcript capture
- the durable byte-accurate tier of the repl verification model recorded in goals/G-044 detail: pure layer (commandcomplete.test, done), console-seam unit tier (G-001 as enabling refactor), rendered tiers (G-020 screen capture near-term windows bridge; this goal headless + cross-platform)
- first verification targets = preserve-list behaviours: the in-proc {set x "{*}{"} continuation-hint sequence, raw-mode tab-marker rendering/deletion, up-arrow recall + edit of a multiline history entry
- detail file records open implementation questions (driving ConPTY from tcl: twapi surface vs helper exe; harness/shell synchronisation incl the inter-subshell beep-protocol idea vs prompt sentinels) and the G-059 WSL / G-060 QEMU-guest execution extensions

Assisted-by: harness=claude; primary-model=claude-fable-5; api-location=anthropic.com
master
Julian Noble 4 days ago
parent
commit
89daa99269
  1. 6
      GOALS.md
  2. 78
      goals/G-061-pseudoconsole-expect.md

6
GOALS.md

@ -409,3 +409,9 @@ Scope: build/test orchestration config and scripts (location to be settled with
Detail: goals/G-060-qemu-test-matrix.md
Goal: comprehensive cross-platform verification (linux variants, FreeBSD/NetBSD/OpenBSD, arm architectures) is achievable from a single dev machine by driving QEMU guests as strictly external tooling - punkshell invokes qemu as a separate process with declarative per-guest config, and nothing in punkshell or its artifact repos bundles, links against, derives from or otherwise couples to QEMU (GPLv2) - so the project's BSD licensing posture is unaffected and QEMU remains a swappable convenience: the guest-driving contract is push-based artifact staging and result collection (the same pattern G-059 establishes for WSL - guests never share a working tree with the host) and must be satisfiable by real hardware, WSL or another hypervisor.
Acceptance: a documented, repeatable workflow provisions at least one non-Windows QEMU guest (e.g FreeBSD x86_64) that fetches a punkbin runtime and runs the source-tree suite inside the guest, driven from the Windows dev machine with results collected back; the guest-driving interface is hypervisor-agnostic (documented contract; QEMU is one provider); the license posture is recorded in the detail file (external-process invocation only, no QEMU binaries or derived code committed to punkshell or punkbin, guest OS images not redistributed by the project - fetched/built per machine); suite runs on machines without QEMU are unaffected.
### G-061 [proposed] Pseudoconsole expect-alternative for interactive shell testing (ConPTY + unix pty)
Scope: test-harness support (location TBD during the work: src/tests/testsupport/ and/or a small punk module), src/tests/shell/ (capability-gated interactive suites), goals/G-061-pseudoconsole-expect.md
Detail: goals/G-061-pseudoconsole-expect.md
Goal: interactive punkshell behaviour is testable headlessly and byte-accurately by driving a real shell under a pseudoconsole - ConPTY on windows, pty on unix - through an expect-like harness (spawn, send literal keys and control/arrow sequences, await patterns in the rendered ANSI byte stream with timeouts, collect transcript), giving the goals/G-044-detail preserve-list behaviours (continuation hints, raw-mode colour staging, tab markers and space dots, multiline history navigation) their durable cross-platform verification tier - complementary to G-020 capture testing (which verifies what reaches the screen; this verifies the byte stream) and to the G-001 console-seam unit tier.
Acceptance: a capability-gated harness spawns a built punkshell under ConPTY on windows (and under a unix pty when run on unix or via the G-059 WSL staging pattern), sends keys including control/arrow sequences, and awaits expected patterns with timeouts and forced teardown on expiry (hang-proof per the shell-test conventions); at least three preserve-list behaviours are verified through it (e.g the in-proc `set x "{*}{"` continuation-hint sequence, raw-mode tab-marker rendering and deletion, up-arrow recall and edit of a multiline history entry); tests skip cleanly where no pseudoconsole is available; the harness API is platform-agnostic with per-OS backends; candidate synchronisation/communication mechanisms between harness and shell (including the inter-subshell beep-protocol idea) are evaluated and the chosen approach documented in the detail file.

78
goals/G-061-pseudoconsole-expect.md

@ -0,0 +1,78 @@
# G-061 Pseudoconsole expect-alternative for interactive shell testing (ConPTY + unix pty)
Status: proposed
Scope: test-harness support (location TBD during the work: src/tests/testsupport/ and/or a small punk module), src/tests/shell/ (capability-gated interactive suites), goals/G-061-pseudoconsole-expect.md
Acceptance: see GOALS.md index entry (canonical).
## Context
punkshell's interactive behaviour has no automated verification. The preserve-list
recorded in goals/G-044-repl-command-completion.md (2026-07-11, user-specified) covers
what any repl refactor must not break: info-complete-parity continuation with the Tcl
quoting quirk, closing-prompt hints, raw-mode colour staging (in-progress vs submitted),
literal tab acceptance with raw-mode tab markers and edit smarts, dim grey space dots
(display-only - never in the submitted string), and up/down navigation + editing of
multiline editbuf history entries.
The verification tier model established 2026-07-11:
1. PURE layer - testable now: punk::lib::system::incomplete characterized in
src/tests/modules/punk/lib/testsuites/lib/commandcomplete.test.
2. Console-seam UNIT tier - blocked on a seam: class_editbuf is console-coupled at its
core (add_chunk renders via overtype::renderline against live terminal metrics -
cursor-position size probing, DECRQPSR tabstops). An ::opunk::Console test double
([[G-001]]) or injectable metrics is the enabling refactor.
3. RENDERED tier: [[G-020]] screencap+input injection is the near-term windows-only
bridge (verifies what reaches the screen); THIS goal is the durable, headless,
cross-platform, byte-accurate mechanism (verifies the byte stream the shell emits).
## Approach
### Harness contract (platform-agnostic; per-OS backends)
expect-like primitives over a REAL pseudo-terminal:
- spawn: launch a built punkshell attached to a pseudoconsole (ConPTY on windows,
openpty/forkpty on unix) with controlled size (cols x rows) - size control matters
because raw-mode rendering is width-dependent.
- send: literal keys plus control/arrow/escape sequences (up-arrow history recall,
backspace over a tab marker, etc).
- expect: await patterns in the rendered ANSI byte stream with timeouts; forced
teardown on expiry (hang-proof, per src/tests/shell/AGENTS.md conventions - the
punk_run pattern generalized to a live conversation).
- transcript: full byte capture for post-hoc assertions (colour staging = SGR
sequences around the in-progress vs submitted line; tab markers/space dots = the
substituted glyph bytes).
### Open implementation questions (deliberately left to the work)
- Driving ConPTY from Tcl: candidates include a twapi surface (if CreatePseudoConsole
is reachable), a small helper executable owning the ConPTY pair and relaying bytes
over stdio/sockets, or an existing tool at arm's length. The unix side is simpler
(openpty via a small C/critcl shim, or driving through an existing pty-capable tool).
- Synchronisation between harness and shell: pattern-awaiting alone can be racy for
timing-sensitive raw-mode behaviour. Candidate: the inter-subshell communication /
beep-protocol idea (user, 2026-07-11) - an in-band signalling channel the shell can
emit for the harness to key off. Evaluate against simpler prompt-sentinel approaches;
document the choice here.
- Unix execution from the windows dev machine: the [[G-059]] WSL staging pattern
(native-filesystem staging, capability constraint) extends naturally - a pty harness
run inside the distro against a linux punkbin runtime/kit.
### Sequencing
- [[G-001]] console-seam unit tier and [[G-020]] capture tier are complementary, not
prerequisites - but preserve-list items verified at a cheaper tier need not be
re-verified here beyond smoke coverage.
- First verification targets (per acceptance): the in-proc `set x "{*}{"`
continuation-hint sequence, raw-mode tab-marker rendering + deletion, up-arrow recall
and edit of a multiline history entry.
## Notes
- Related: [[G-044]] (completion/hinting - consumes this for its own verification),
[[G-001]] (fake-console unit tier), [[G-020]] (screen-capture tier), [[G-059]] (WSL
execution pattern), [[G-060]] (guest-driving contract - a pty harness inside a QEMU
guest is the eventual cross-platform matrix combination).
- ConPTY availability gates the windows side to Win10 1809+ (fine for punkshell's
supported baseline); the capability probe should verify actual ConPTY function, not
windows version.
Loading…
Cancel
Save