perf(web): defer pull request line stats until visible

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}

[Adamulek123]

Pull request line counts used to load for every row, including rows far outside the viewport. Large lists could send hundreds of expensive host stat reads before the user reached those rows.

This change uses one list-owned observer to request stats for rows near the viewport. It groups stable requests by environment, keeps received counts cached, and does not add active or cached rows to later batches.

Largest and smallest sorting still requests stats for every loaded row because those counts decide the row order. Manual refresh uses the same policy. Each environment batch stays within the 500-ref contract limit. If the filter or sort policy changes while a manual refresh waits for host invalidation, the stale stats refresh is ignored.

This modernizes @Adamulek123's original implementation on current main while preserving the new row layout, filters, facet refresh, scroll behavior, and sorting.

Validation

• vp test run apps/web/src/components/pullRequest/pullRequestList.logic.test.ts (102 tests passed)
• vp run --filter @t3tools/web typecheck
• Targeted type-aware lint and format checks for the three policy source and test files
• git diff --check origin/main...HEAD

No browser test was run for this logic and request-lifetime change.

Modernized by GPT-5.6 Sol in the Codex harness. Original implementation by @Adamulek123.

---

Note

Medium Risk
Touches request lifetime, refresh races, and sort-dependent data loading on a high-traffic list; logic is heavily unit-tested but browser behavior was not validated.

Overview
Pull request diff line counts are no longer fetched for every loaded row. The list now drives stats through a shared IntersectionObserver on the scroll container, with each PullRequestRow registering via statsRef and data-pull-request-stats-key.

New logic in pullRequestList.logic batches refs per environment (max 500 per batch), skips rows already in-flight or cached, prunes completed batches when rows scroll away, and chooses visible vs eager policy from sort mode (largest/smallest still loads all rows). Manual refresh re-requests only under the current filter/policy scope.

usePullRequestListStats / merged environment queries gain isPending and targeted refresh(override) so refresh does not blindly re-hit every prior batch. Pagination and header observers now share the parent scrollRef instead of the viewport.

^{Reviewed by Cursor Bugbot for commit 18e9cdf. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Defer pull request diff-stat fetches until rows are visible

• Diff-stat requests for the pull-requests list are now batched by environment (max 500 refs per batch) and deferred until rows enter the viewport, driven by an IntersectionObserver on each row button.
• Sorting by size uses an eager policy that requests stats for all loaded rows; other sorts use a visible policy that only requests stats for rows currently on screen.
• Refresh triggers scoped re-fetches via pullRequestStatsRefreshBatches, ignoring late refreshes when the filter or policy has changed; completed batches whose rows have left the viewport are pruned with retainVisiblePullRequestStatsBatches.
• PullRequestRow gains statsKey/statsRef props and a data-pull-request-stats-key attribute; PullRequestsColumn now receives a shared scrollRef so list paging and stats observers share one scroll root.
• usePullRequestListStats now exposes isPending and accepts an explicit targets override on refresh.
• Behavioral Change: size-based sorts now request stats for the full loaded list rather than only visible rows; the stats observer shares the scrollRef root with the list paging observer, so any consumer overriding PullRequestsColumn's scroll container must pass a valid ref.

^{Macroscope summarized 18e9cdf.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 2578675f-6998-4024-88c9-10712ae67a9e

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This changes the default pull-request list’s stats-loading behavior and adds substantial observer, batching, caching, refresh, and sort-dependent request lifecycle logic. The interaction between viewport visibility, pagination, sorting, and refresh warrants human review.

^{You can add or adjust custom eligibility rules. Learn more.}

[macroscopeapp]

One finding on the new visibility-driven stats observer: the rootMargin lookahead is inert because the observer uses the implicit viewport root while the list scrolls inside the page's own overflow-y-auto container, so line counts land after rows are already on screen and reflow the row's right column in view.

^{Posted via Macroscope — UI Consistency}

Dismissing prior approval to re-evaluate 61e36cb

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 27f2176. Configure here.}