Land mainstream-usability P0-P5 on the v0.2 main line - #58
Open
Hyper66666 wants to merge 19 commits into
Open
Conversation
One TLS connection carries a routed buffered answer, a chunked streaming response whose finish() keeps the connection alive, and a routed Connection: close answer (tls_composes_with_router_keep_alive_and_streaming, Windows Schannel). Locked realworld loop re-run green; matrix and network-runtime constraints updated to current behavior. Closes http-production-serving tasks 5.4 and 7.4. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Six priorities from verified probes: P0 Option/Result as enums (root cause: struct-shaped optionals force sentinel payloads), P1 for over collections, P2 quiet output and one diagnostic language, P3 everyday syntax, P4 idiomatic flagship examples, P5 width-aware sgfmt blocks. Records the sgfmt collapse root cause (max_width parsed but never read) and the pre-existing sgpm v1 lockfile ordering defect as D7/D8. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Successful build/run prints only program output plus one result line; cache, workset, frontend session/scheduler, generic-instance counters, and pass-through clang include-path warnings move behind --verbose. Chinese typeck messages (undefined variable/type, unknown method, argument count) become English with stable codes unchanged. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
format_block_inline falls back to one-statement-per-line rendering when the inline form would exceed max_width, activating the parsed-but-unread option for every block form (if/while/for/loop/match arms/async/ parallel/try). Short blocks stay inline byte-identical; width budget counts indent and header prefix. 18 formatter tests cover breaking, nesting, else-if chains, and idempotence. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
load_config drops from one 700-char five-deep line to guard clauses with early return plus extracted helpers; behavior unchanged, locked loop green (check, 3 tests, fmt). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Generic enums monomorphise per instance: MIRType::Enum carries the instance name, EnumDef keeps HIR payload shapes and instantiates them with a recursion guard. Payload-free variants infer their arguments from the expected type (annotated let, return position) exactly like struct literals. Bare Some/None/Ok/Err resolve to their unique declaring enum in expressions and match patterns, with ambiguous-enum-variant when two enums share a name; lexical scope still wins. Exhaustiveness treats a bare variant arm as that variant, not a catch-all. Single-element payloads bind whole in match arms (a String payload no longer splits into word fields). The ? operator propagates enum-shaped Result and Option unchanged. Six lowering regression tests pin these paths; compiler suite 1122 green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A v1 lockfile stores source as a string where v2 expects a table, so deserializing the v2 shape first rejected every v1 lock as 'out of date' while update --check called the same file current. The version is now probed from a toml::Value first: non-v2 documents return the graceful empty pin set the existing check intended, unreadable files report the real IO error, and unparseable documents report 'malformed' with the file named instead of a false staleness claim. Regression tests cover v1 tolerance, malformed rejection, and the missing-file path; the stale-garbage integration test now expects the malformed wording. 5 of the 7 previously failing fixtures pass check --locked; p0-foundations (no lockfile committed) and package-release-loop (checksum-less registry entry) fail for real, correctly named reasons. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mechanical sgfmt -w over the ten files the new max_width rule breaks differently; workspace-doc-loop/src/lib.sg expands 18 -> 115 lines, showing how much structure the inline form had collapsed. Behavior unchanged; sgpm integration and realworld_e2e suites green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Keep array, slice, and range direct lowering, and add map keys()/values() adapters. Co-authored-by: Cursor <cursoragent@cursor.com>
Keep compiler instrumentation behind --verbose and lock the existing error JSON shape. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Fix for-loop None matching, BTreeMap keys() LLVM types, and the realworld fmt/e2e follow-ups so the section 8 cargo and OpenSpec gates pass. Co-authored-by: Cursor <cursoragent@cursor.com>
All P0-P5 tasks and section 8 verification gates are done, so move the change into the dated archive. Co-authored-by: Cursor <cursoragent@cursor.com>
…ility-p0-p5 Co-authored-by: Cursor <cursoragent@cursor.com> # Conflicts: # docs/network-runtime.md # examples/realworld/SUPPORT_MATRIX.md # examples/realworld/http-echo-service/src/main.sg # examples/realworld/http-echo-service/tests/http_echo_smoke.sg # openspec/changes/archive/2026-07-25-http-production-serving/tasks.md # runtime/src/net/http_server.rs # tools/sgc/src/cli.rs # tools/sgc/src/commands/run.rs # tools/stdlib/net.sg
The lambda owned-param drop helper was missing the drop-flag fields main added, so the compiler crate would not build. Co-authored-by: Cursor <cursoragent@cursor.com>
Main's new parent-drop assertion treated any String_Drop_drop as a leak; flagged glue still emits the call, so require a live-flag guard instead. Co-authored-by: Cursor <cursoragent@cursor.com>
5 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
mainstream-usability-p0-p5on currentmain: enumOption/Result,forover collections, quiet Englishsgcdiagnostics,vec!/ format print /{:?}/if let, flagship guard-clause rewrites, and width-awaresgfmt.origin/main(v0.2 release evidence) into this branch. Keep--verboseplus--runtime-mode, keep main's Vec-based HTTP router /serve_http_timeout, and convert newResult { ... }struct literals back toOk/Err.DropBindinglive-flag fields after the merge and accept flagged (not unguarded) parent drop glue when bothif/elsearms move an owned aggregate.Test plan
cargo test -p sengoo-compiler --lib(1173 passed locally after the merge fixes)cargo test -p sgc --test quiet_output— check/json cases passed;sgc runfailed locally with missing MSVC headers (environment), not an assertion on quiet outputsgc,sgpm,sglsp,sgfmt, clippy-D warningshttp-echo-serviceafter the router/Ok/Errrewrite