Skip to content

fix(tests): drain three base-reds left by the SSE-comment default and a locale gap - #10704

Merged
diegosouzapw merged 2 commits into
release/v3.8.50from
fix/release-v3.8.50-basereds-sse-vi
Aug 19, 2026
Merged

diegosouzapw merged 2 commits into
release/v3.8.50from
fix/release-v3.8.50-basereds-sse-vi

Conversation

@diegosouzapw

Copy link
Copy Markdown
Owner

Three reds that reproduce on a pristine checkout of the release tip, with zero local changes.
None is caused by the branch that surfaced them (a docs/Dockerfile PR, #10701, whose full unit
run exposed them).

1. i18n-vi-completeness — 3 tests

Six keys landed in en.json with no Vietnamese counterpart:
settings.reasoningTokenBuffer(Desc), settings.zeroLatencyOptimizations(Desc),
settings.compressionOutputStyle.i-have-adhd.{label,description}. The vi locale is held to
strict parity, so the whole suite went red.

Translated. No existing key reordered — the diff is 9 added lines.

2. chatcore-translation-paths — "emits final SSE metadata comments before [DONE]"

#10539 flipped OMNIROUTE_SSE_COMMENTS to off-by-default (strict SSE clients JSON.parse every
line and crash on : comments) and updated three sibling tests — but not this one, which
asserted the trailer is emitted.

The test now asserts the current contract: the stream is comment-free, still ends with [DONE],
and the metadata still reaches the client through the X-OmniRoute-* headers. The opt-in half —
trailer present, after the finish chunk, before [DONE] — stays covered by
sse-comments-optout-9305.test.ts, which drives the env var through all three states.

First attempt was to set OMNIROUTE_SSE_COMMENTS=on inside the test. That turned one red into
a different one: the env var leaks into the sibling call-log tests in the same file (chatCore records Claude prompt cache… started failing, and passes in isolation). Reverted.

3. chat-messages-validation-6402 — all 9 Antigravity cases

These asserted assert.match(body, /ok/) against the mocked model output. That text never
reached this layer.
The match only ever succeeded on the ok inside
: x-omniroute-tokens-in=0 — an SSE comment trailer. Dumping the body at the last passing
commit shows it contained only trailers, no model text:

": x-omniroute-cache-hit=false\n: x-omniroute-latency-ms=26\n…: x-omniroute-tokens-in=0\n…"

When the trailers stopped being emitted, the coincidence broke — not the behavior. Bisected to
6b823aa441 (its parent 6d99a46d4b passes); forcing OMNIROUTE_SSE_COMMENTS=on at the tip
reproduces the trailers-only body, confirming nothing about content relay changed.

The test now asserts the guard it is named for: a cloudcode envelope (no messages key) must
not be rejected by the #6402 missing-messages validator. Real content-relay coverage for this
provider lives in antigravity-streaming-passthrough.test.ts (passes, asserts
choices[0].message.content === "Hello again").

Verification

Suite Before After
i18n-vi-completeness 2 pass / 3 fail 5 pass / 0 fail
chatcore-translation-paths 69 / 1 70 / 0
chat-messages-validation-6402 5 / 9 14 / 0

Base-reds still open (not in this PR)

Seven more files reproduce on the same tip and need their own triage — some may be product
regressions rather than stale tests:

antigravity-model-aliases (3) · glm-provider-model-import-route (7) · executor-qwen-web (1) ·
hard-session-lease-bypass-inventory (1) · i18n-glossary-consistency-check (1, zh-TW) ·
instrumentation-warm-catalog-cache (2) · model-sync-route (2)

Plus npm run lint: 25 errors in tests/unit/cli-oauth-commands.test.ts and
tests/unit/executor-gitlab.test.ts.

Inventory posted to #9985. Note that Release-Green reported "Validate active release branch:
success"
while all of these were red, so that signal is not covering the unit suite the PR
checks run.

Refs #9985, #10539

… a locale gap

All three reproduce on a pristine tip; none is caused by the branch that found
them.

1. i18n vi — six keys landed in en.json without a Vietnamese counterpart
   (settings.reasoningTokenBuffer*, settings.zeroLatencyOptimizations*,
   settings.compressionOutputStyle.i-have-adhd.*). The vi locale is held to
   strict parity, so the whole i18n-vi suite went red. Translated; no existing
   key reordered.

2. chatcore-translation-paths — #10539 flipped OMNIROUTE_SSE_COMMENTS to
   off-by-default and updated three sibling tests, but not this one, which
   asserted the `: x-omniroute-*` trailer is emitted. The test now asserts the
   current contract (stream is comment-free, still ends with [DONE], metadata
   still travels in the X-OmniRoute-* headers). The opt-in half stays covered by
   sse-comments-optout-9305.test.ts, which drives the env var through all three
   states. Enabling the flag inside this file instead leaks process.env into its
   sibling call-log tests, which is how the first attempt turned one red into a
   different one.

3. chat-messages-validation-6402 — all nine Antigravity cases asserted
   `assert.match(body, /ok/)` against the mocked model output. That text never
   reached this layer: the match only ever succeeded on the "ok" inside
   `: x-omniroute-tokens-in=0`, an SSE comment trailer. When the trailers stopped
   being emitted the coincidence broke, not the behavior — bisected to
   6b823aa, whose parent 6d99a46 passes. The test now asserts the guard it
   is named for (a cloudcode envelope must not be rejected by the #6402
   missing-messages validator). Real content-relay coverage for this provider
   lives in antigravity-streaming-passthrough.test.ts, which passes.

Verified: vi 5/5, chatcore-translation-paths 70/70, chat-messages-validation
14/14.
@diegosouzapw
diegosouzapw merged commit c8d531d into release/v3.8.50 Aug 19, 2026
11 checks passed
@diegosouzapw
diegosouzapw deleted the fix/release-v3.8.50-basereds-sse-vi branch August 19, 2026 15:10
giauphan pushed a commit to giauphan/OmniRoute that referenced this pull request Aug 20, 2026
… a locale gap (diegosouzapw#10704)

* fix(tests): drain three base-reds left by the SSE-comment default and a locale gap

All three reproduce on a pristine tip; none is caused by the branch that found
them.

1. i18n vi — six keys landed in en.json without a Vietnamese counterpart
   (settings.reasoningTokenBuffer*, settings.zeroLatencyOptimizations*,
   settings.compressionOutputStyle.i-have-adhd.*). The vi locale is held to
   strict parity, so the whole i18n-vi suite went red. Translated; no existing
   key reordered.

2. chatcore-translation-paths — diegosouzapw#10539 flipped OMNIROUTE_SSE_COMMENTS to
   off-by-default and updated three sibling tests, but not this one, which
   asserted the `: x-omniroute-*` trailer is emitted. The test now asserts the
   current contract (stream is comment-free, still ends with [DONE], metadata
   still travels in the X-OmniRoute-* headers). The opt-in half stays covered by
   sse-comments-optout-9305.test.ts, which drives the env var through all three
   states. Enabling the flag inside this file instead leaks process.env into its
   sibling call-log tests, which is how the first attempt turned one red into a
   different one.

3. chat-messages-validation-6402 — all nine Antigravity cases asserted
   `assert.match(body, /ok/)` against the mocked model output. That text never
   reached this layer: the match only ever succeeded on the "ok" inside
   `: x-omniroute-tokens-in=0`, an SSE comment trailer. When the trailers stopped
   being emitted the coincidence broke, not the behavior — bisected to
   6b823aa, whose parent 6d99a46 passes. The test now asserts the guard it
   is named for (a cloudcode envelope must not be rejected by the diegosouzapw#6402
   missing-messages validator). Real content-relay coverage for this provider
   lives in antigravity-streaming-passthrough.test.ts, which passes.

Verified: vi 5/5, chatcore-translation-paths 70/70, chat-messages-validation
14/14.

* chore(changelog): correct the fragment to the real PR number (diegosouzapw#10704)

---------

Co-authored-by: Xiangzhe <bakryun0718@proton.me>
muhamadgalihsaputra pushed a commit to niyatna/NiyatnaRoute that referenced this pull request Sep 27, 2026
… a locale gap (diegosouzapw#10704)

* fix(tests): drain three base-reds left by the SSE-comment default and a locale gap

All three reproduce on a pristine tip; none is caused by the branch that found
them.

1. i18n vi — six keys landed in en.json without a Vietnamese counterpart
   (settings.reasoningTokenBuffer*, settings.zeroLatencyOptimizations*,
   settings.compressionOutputStyle.i-have-adhd.*). The vi locale is held to
   strict parity, so the whole i18n-vi suite went red. Translated; no existing
   key reordered.

2. chatcore-translation-paths — diegosouzapw#10539 flipped OMNIROUTE_SSE_COMMENTS to
   off-by-default and updated three sibling tests, but not this one, which
   asserted the `: x-omniroute-*` trailer is emitted. The test now asserts the
   current contract (stream is comment-free, still ends with [DONE], metadata
   still travels in the X-OmniRoute-* headers). The opt-in half stays covered by
   sse-comments-optout-9305.test.ts, which drives the env var through all three
   states. Enabling the flag inside this file instead leaks process.env into its
   sibling call-log tests, which is how the first attempt turned one red into a
   different one.

3. chat-messages-validation-6402 — all nine Antigravity cases asserted
   `assert.match(body, /ok/)` against the mocked model output. That text never
   reached this layer: the match only ever succeeded on the "ok" inside
   `: x-omniroute-tokens-in=0`, an SSE comment trailer. When the trailers stopped
   being emitted the coincidence broke, not the behavior — bisected to
   d5c4a54, whose parent 639f810 passes. The test now asserts the guard it
   is named for (a cloudcode envelope must not be rejected by the diegosouzapw#6402
   missing-messages validator). Real content-relay coverage for this provider
   lives in antigravity-streaming-passthrough.test.ts, which passes.

Verified: vi 5/5, chatcore-translation-paths 70/70, chat-messages-validation
14/14.

* chore(changelog): correct the fragment to the real PR number (diegosouzapw#10704)

---------

Co-authored-by: Xiangzhe <bakryun0718@proton.me>
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.

2 participants