ENG-9611: defer tracking/analytics scripts so they don't block render by adhami3310 · Pull Request #6568 · reflex-dev/reflex

adhami3310 · 2026-05-27T19:34:35Z

Summary

External src tracker scripts (gtag loader, clearbit) get async_=True so they no longer block HTML parsing while downloading.
Inline tracker bootstraps (posthog, koala, common_room, rb2b, unify, default, gtag init) get type="module", which the browser defers until after HTML parsing.
All eight telemetry script components switch from rx.script to rx.el.script. These trackers reach the DOM via App(head_components=...), which reflex/compiler/utils.py:create_document_root compiles straight into <Head> — so the Helmet wrapper that rx.script adds is redundant reconciliation cost.
gtag_report_conversion is intentionally left non-module: it defines a global function gtag_report_conversion() that page onclick handlers call by name; module scope would hide it from window.

Cross-repo follow-up

get_common_room_trackers is currently rendered inside rx.theme(rx.cond(...)) in two consumers (cloud, flexgen) rather than head_components. The old rx.script/Helmet wrapper hoisted it to <head> regardless; switching to rx.el.script means it now stays in the body where the theme renders it. The IIFE is idempotent and type="module" defers the same way regardless of DOM position, so it still works — but the head/body location changes for those two consumers.

A follow-up PR to cloud and flexgen should move the get_common_room_trackers(...) call into head_components= and adapt the rx.cond(~AuthState.in_session & is_prod_mode(), ...) gate accordingly.

Test plan

uv run ruff check . clean
uv run ruff format . clean
uv run pyright clean on the telemetry package
After merge/deploy: open DevTools → Network on a site using get_pixel_website_trackers() and confirm tracker requests are no longer in the render-blocking section of the waterfall

External `src` trackers get `async_=True` (gtag loader, clearbit). Inline trackers get `type="module"` so the browser defers them until after HTML parsing (posthog, koala, common_room, rb2b, unify, default, gtag init). Also switches all of these from `rx.script` to `rx.el.script`. The trackers are placed via `App(head_components=...)` which is compiled directly into `<Head>`, so the Helmet wrapper that `rx.script` adds is redundant runtime cost. `gtag_report_conversion` is intentionally left non-module — it defines a global function called by name from `onclick` handlers.

linear-code · 2026-05-27T19:34:38Z

ENG-9611

codspeed-hq · 2026-05-27T19:37:54Z

Merging this PR will not alter performance

✅ 24 untouched benchmarks

_{Comparing khaleel/eng-9611-defer-tracker-scripts (9e37f01) with main (cf5e4ed)¹}

No successful run was found on main (8ad0f56) during the generation of this report, so cf5e4ed was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

greptile-apps · 2026-05-27T19:39:12Z

Greptile Summary

This PR defers all telemetry/analytics scripts to eliminate render-blocking by applying async_=True to external src scripts (Clearbit, gtag loader) and type="module" to inline bootstrap scripts (PostHog, Koala, Common Room, RB2B, Unify, Default, and the gtag init snippet). All eight tracker components are also migrated from rx.script (Helmet wrapper) to rx.el.script (raw element) since they are placed directly into <head> via head_components.

The previous review concern about function gtag() being module-scoped and therefore invisible to window has been addressed: the bootstrap now uses window.gtag = window.gtag || function gtag() {...} to explicitly assign to the global object.
Every other inline IIFE was already module-safe: all globals they expose (window.posthog, window.reb2b, window.ko, window.signals, window.unify, window.__default__) are assigned via explicit window.* properties.
gtag_report_conversion intentionally remains a plain (non-module) rx.el.script so that the function gtag_report_conversion() declaration lands on window and is reachable by onclick handlers, and so its internal bare gtag() call can resolve window.gtag from the global scope.

Confidence Score: 5/5

Changes are safe to merge; all inline scripts correctly expose their globals via explicit window.* assignments, and the intentional non-module exception for gtag_report_conversion is well-reasoned.

The previous concern about window.gtag not being set from within a module script was the key risk here, and it has been correctly resolved by explicitly assigning window.gtag. Every other converted script already used window.* for any globals it exposes, so the type="module" deferral is safe across the board.

No files require special attention. The one file worth a second glance is google.py for the window.gtag fix and the intentional non-module treatment of gtag_report_conversion, both of which are handled correctly.

Important Files Changed

Filename	Overview
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/google.py	GTAG init bootstrap now uses `window.gtag = window.gtag
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/posthog.py	Inline PostHog bootstrap converted to `rx.el.script` with `type="module"`; IIFE assigns to `window.posthog` explicitly, and bare `posthog.init()` call at module top-level correctly resolves from global scope.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/rb2b.py	Both RB2B pixel scripts switched to `rx.el.script` with `type="module"`; both IIFEs assign to `window.reb2b` explicitly, making them module-safe.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/clearbit.py	External Clearbit src script correctly gets `async_=True` to stop blocking HTML parsing.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/common_room.py	Converted to `rx.el.script` with `type="module"`; IIFE uses `window.signals` assignment, so the global is properly accessible after deferral.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/koala.py	Converted to `rx.el.script` with `type="module"`; IIFE sets `window.ko` explicitly before using the bare `ko` identifier, which resolves correctly from the global scope in a module context.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/default.py	Converted to `type="module"`; the minified IIFE uses named-function-expression idiom (not declaration), so `window.__default__` assignment via the `e` parameter is unaffected by module scope.
packages/reflex-components-internal/src/reflex_components_internal/blocks/telemetry/unify.py	Converted to `rx.el.script` with `type="module"`; IIFE assigns to `window.unify` and `window.unifyBrowser` explicitly, compatible with module scope.

_{Reviews (2): Last reviewed commit: "fix: assign gtag to window so type=modul..." | Re-trigger Greptile}

A top-level `function gtag(){}` in a type=module script is module-scoped, not installed on `window`. That breaks `gtag_report_conversion` and any page code that calls `gtag('event', ...)` before the external gtag.js loader finishes — `window.gtag` is undefined and the call throws. Assign `window.gtag` explicitly. The `|| window.gtag` guard handles the case where the async external loader executes before the deferred module inline (so we don't clobber gtag.js's real implementation). Caught by Greptile on PR #6568.

adhami3310 · 2026-05-27T19:54:43Z

@greptile

adhami3310 requested a review from a team as a code owner May 27, 2026 19:34

masenf approved these changes May 28, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ENG-9611: defer tracking/analytics scripts so they don't block render#6568

ENG-9611: defer tracking/analytics scripts so they don't block render#6568
adhami3310 wants to merge 2 commits into
mainfrom
khaleel/eng-9611-defer-tracker-scripts

adhami3310 commented May 27, 2026

Uh oh!

linear-code Bot commented May 27, 2026

Uh oh!

codspeed-hq Bot commented May 27, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot commented May 27, 2026 •

edited

Loading

Uh oh!

adhami3310 commented May 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

adhami3310 commented May 27, 2026

Summary

Cross-repo follow-up

Test plan

Uh oh!

linear-code Bot commented May 27, 2026

Uh oh!

codspeed-hq Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merging this PR will not alter performance

Footnotes

Uh oh!

greptile-apps Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Important Files Changed

Uh oh!

adhami3310 commented May 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codspeed-hq Bot commented May 27, 2026 •

edited

Loading

greptile-apps Bot commented May 27, 2026 •

edited

Loading