Stream G: run the front end once per Sky.Spa build
Machine: 16 GB Mac. Release builds. Median of 3 warm runs. Parent = the sky build
process (its own SKY_TIMINGS table). base = perf/build-speed HEAD c4d17f467.
Whole build, N=200
| app | path | front half (parent) | split step | build wall | parent peak RSS | front-end note |
|---|---|---|---|---|---|---|
| genwhole 200, direct Spa entry | base | 0.65 s | 0.62 s | 4.70 s | 328 MB | 2 loads, 2 checks |
| genwhole 200, direct Spa entry | new | 0.66 s | 0.15 s | 4.20 s | 330 MB | 1 load, 1 check |
| bench-gen 200, direct Spa entry | base | 0.28 s | 0.19 s | 1.54 s | 155 MB | 2 loads, 2 checks |
| bench-gen 200, direct Spa entry | new | 0.28 s | 0.05 s | 1.45 s | 153 MB | 1 load, 1 check |
| genwhole 200, --target web:app | base | 0.67 s | 0.62 s | 4.78 s | 328 MB | 2 loads, 2 checks |
| genwhole 200, --target web:app | new | 0.66 s | 0.62 s | 4.79 s | 329 MB | 2 loads, 2 checks |
| bench-gen 200, --target web:app | base | 0.28 s | 0.20 s | 1.62 s | 151 MB | 2 loads, 2 checks |
| bench-gen 200, --target web:app | new | 0.28 s | 0.20 s | 1.63 s | 153 MB | 2 loads, 2 checks |
"Direct Spa entry" = the staged .skyapp/web-app project (the synthesised
Spa.app entry) built as its own project. Base counts are by reading the code
(the base binary has no counter). Neither generated app keeps a continuation on
the server, so the shared second check does not show in these rows; the
fixture test shows it (2 checks -> 1).
An earlier cut held the front half's CheckOutput.def_types alive through the
split. The web:app parent peak then rose to 385-396 MB in most runs (a control
binary with no behaviour change measured 327-331 MB). The per-def types are now
dropped before the split (no analysis reads them); peak is back to 329 MB.
Byte identity
sky spa-splitoutput, base vs new: identical for genwhole 100, genwhole 200, bench-gen 200, spa-split-multimodule fixture, and both direct-entry projects (10, 10, 20, 10, 10, 20 files).sky build(reused db) split vs basesky spa-split: identical for direct genwhole 200, direct bench-gen 200, spa-split-multimodule.
Diagnostics
111 runs, base vs new, stdout + stderr + exit, durationMs masked: 0 differ.
- reject corpus (89 files):
sky check --format json. - spa-split-multimodule variants (clean, client-arm, server-arm, shared module,
server module, two errors, name error):
sky check --format jsonandsky build --format json. - 62-app-notes variants (clean, client-arm, server-arm, shared module, server
module):
sky check/build --target web:app --format json.
Tests
- project
spa_split_checks_once(new): red (1 load, 2 checks) -> green (1, 1). - sky
spa_split_flow::a_direct_spa_build_runs_the_front_end_once(new): red (sky check3 loads, 3 checks) -> green (check 2/2, build 1/1, split byte-identical to a freshsky spa-split). - project lib
spa(59),timings(2),spa_server_chain(10): pass. - spa_split_flow diagnostic and refusal tests (16 + the new one): pass.