Skip to content

Fix: Write the smoke host's measure requests at the record's real length (PR-2.1) - #809

Merged
b-re-w merged 2 commits into
developfrom
fix/smoke-host-measure-record
Nov 14, 2026
Merged

b-re-w merged 2 commits into
developfrom
fix/smoke-host-measure-record

Conversation

@b-re-w

@b-re-w b-re-w commented Nov 14, 2026

Copy link
Copy Markdown
Member

Fixes the renderer smoke test failing on every platform in develop CI run 43617787436 with "the measure round trip did not agree, or never ran" (iOS simulator, GraalVM Linux x64/arm64 and macOS, Windows).

Cause. #788 (FR-83, font features in Host measure) grew the measure request record from 72 to 80 bytes. renderer/desktop/c/smoke_host.c writes its requests by hand and kept 72, so the renderer could not read them. It is the smoke host on every platform.

Fix. MEASURE_RECORD is 80; the records are zeroed first, so the font features slot reads as none. scripts/tests/smoke-host-measure-record.test.sh holds the C length equal to MEASURE_RECORD_LEN (fails on 72).

Proof, GraalVM macOS renderer built from this branch, same dylib both times:

80: record 0: 0 154.000x20.000 / record 1: 0 154.000x20.000 / record 2: 0 166.000x19.000 / record 3: 0 166.000x19.000 -> agree, smoke-test.sh exit 0
72: every record: 2 0.000x0.000 -> disagree, "the measure round trip did not agree, or never ran", exit 1

Second commit. fork-dependencies.sh asked for a per-target copy of skiko-awt, which is one JVM jar with no per-target publications, so require-fork-artifacts.sh desktop refused a complete local Maven repository and no renderer build script could start. Same commit as in #808, brought forward because local renderer builds on develop need it now; its test case fails before the fix.

Not run here: the Linux, Windows and iOS smoke tests (CI).

…gth (PR-2.1)

The measure request record grew from 72 to 80 bytes when the font features
went into it (#788, FR-83). The smoke host writes its requests by hand in C and
kept 72, so the renderer read the second, third and fourth request of the round
trip eight, sixteen and twenty-four bytes early, and the smoke test failed on
every platform with "the measure round trip did not agree, or never ran"
(develop CI run 43617787436: iOS simulator, GraalVM Linux and macOS, Windows).

MEASURE_RECORD is 80. The records are zeroed before they are filled, so the new
font features slot reads as none. scripts/tests/smoke-host-measure-record.test.sh
holds the C length equal to MEASURE_RECORD_LEN; it fails on the old value.
Since 1.11.1-ext.4 (fork #167) the fork's desktop modules (ui-desktop,
foundation-desktop and the rest) name org.thisisthepy.compose.skiko:
skiko-awt at the fork version. skiko-awt is a JVM jar published as a
root alone, with no per-target publications, and publish-skiko-awt.sh
puts it in ~/.m2. fork-dependencies.sh still asked every root sibling
for a publication per served target (skiko-awt-desktop), so
require-fork-artifacts.sh desktop refused a complete repository and
every renderer build script stopped before the toolchain ran.

A root whose metadata has no variant pointing at a target publication
("available-at") is now published when its own files are. A
multiplatform root without the target's publication is still refused.
fork-dependencies.test.sh covers both; the first case failed before
this change with skiko-awt named as missing.
@b-re-w
b-re-w merged commit ddc2777 into develop Nov 14, 2026
@b-re-w
b-re-w deleted the fix/smoke-host-measure-record branch November 14, 2026 18:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant