preflight under 20 minutes, CI under 35 per job, and table keys hash the same on every rail - #3939

Merged

borisbat merged 17 commits into

masterfrom

bbatkin/preflight-20min

Sep 5, 2026

Commits on Sep 5, 2026

preflight: tiers (fast / lane / module), reach sets, concurrent lanes, compile-sweep and utils-tests gates; make-pr's default chain slims

borisbat
and
claude
committed
5bb1219
preflight: the compile sweep splits engine roots onto one serial module cache, sizes its pool by cores, and excludes fixture trees

borisbat
and
claude
committed
0d44741
preflight: lane verdicts parse behind the to_log level prefix, and the AOT gate names its failing files

borisbat
and
claude
committed
f8447d0
CI: the job matrices become data, extended_checks splits into two darwin roles per PR, and the over-budget cells move to the nightly

borisbat
and
claude
committed
65c6f0c
tables: a vector or range key hashes the same bytes on every rail, so a table keeps its keys past the first grow

borisbat
and
claude
committed
9996f41
preflight: a ci-matrix gate runs the workflow matrix test, and the workflows checklist states the budget move as three rules

borisbat
and
claude
committed
4e51770
CI: CodeQL stays per PR, the matrix-test rule moves to ci/REVIEW.md, and a ci/REVIEW.das gate wires every ci/test_*.py

borisbat
and
claude
committed
b973a1f
CI: the workflows checklist binds checks, not steps, and the two platform-bound nightly steps say so

borisbat
and
claude
committed
d869aff
preflight: the pure predicates move beside their tests, the matrix test pins the nightly-only steps, and the checklists say what weakening means

borisbat
and
claude
committed
07de6ca
tables: KeyHash tells a builtin key from a handled one instead of comparing byte widths

borisbat
and
claude
committed
af7f764
review: the workflows, ci and simulate checklists after the exit pass - one concept per rule, no positional cites

borisbat
and
claude
committed
a0b4a1a
tests-cpp: KeyHash agrees with hash_function for every builtin key type and takes the workhorse detour for a handled one

borisbat
and
claude
committed
53eecca
module cache: the default directory is capped - the oldest records go past DAS_MODULE_CACHE_LIMIT megabytes

borisbat
and
claude
committed
d995b23
module cache: the eviction helpers compile only with file IO

borisbat
and
claude
committed
5dd1fd3
tables: WrapsBuiltinValue's marks live beside the WrapType primary in cast.h

borisbat
and
claude
committed
7c5ef2a
CI: the modules job builds the daslang grammar before the MCP tools test

borisbat
and
claude
committed
199f144
CI: the news-region step installs markdown into the interpreter that runs it

borisbat
and
claude
committed
fe1c9a8

Navigation Menu

Uh oh!

preflight under 20 minutes, CI under 35 per job, and table keys hash the same on every rail - #3939

preflight under 20 minutes, CI under 35 per job, and table keys hash the same on every rail#3939
borisbat merged 17 commits into
masterfrom
bbatkin/preflight-20min

Commits on Sep 5, 2026

preflight: tiers (fast / lane / module), reach sets, concurrent lanes, compile-sweep and utils-tests gates; make-pr's default chain slims

preflight: the compile sweep splits engine roots onto one serial module cache, sizes its pool by cores, and excludes fixture trees

preflight: lane verdicts parse behind the to_log level prefix, and the AOT gate names its failing files

CI: the job matrices become data, extended_checks splits into two darwin roles per PR, and the over-budget cells move to the nightly

tables: a vector or range key hashes the same bytes on every rail, so a table keeps its keys past the first grow

preflight: a ci-matrix gate runs the workflow matrix test, and the workflows checklist states the budget move as three rules

CI: CodeQL stays per PR, the matrix-test rule moves to ci/REVIEW.md, and a ci/REVIEW.das gate wires every ci/test_*.py

CI: the workflows checklist binds checks, not steps, and the two platform-bound nightly steps say so

preflight: the pure predicates move beside their tests, the matrix test pins the nightly-only steps, and the checklists say what weakening means

tables: KeyHash tells a builtin key from a handled one instead of comparing byte widths

review: the workflows, ci and simulate checklists after the exit pass - one concept per rule, no positional cites

tests-cpp: KeyHash agrees with hash_function for every builtin key type and takes the workhorse detour for a handled one

module cache: the default directory is capped - the oldest records go past DAS_MODULE_CACHE_LIMIT megabytes

module cache: the eviction helpers compile only with file IO

tables: WrapsBuiltinValue's marks live beside the WrapType primary in cast.h

CI: the modules job builds the daslang grammar before the MCP tools test

CI: the news-region step installs markdown into the interpreter that runs it