Encode telemetry IDs as UUID strings to prevent PostHog truncation

[masenf]

Type of change

• Bug fix (non-breaking change which fixes an issue)

Description

Fixes a critical telemetry bug where 128-bit installation and project identifiers were being silently truncated by PostHog when sent as raw JSON integers. PostHog coerces large numbers to floats, discarding all but ~16 significant digits, causing distinct installations and apps to collapse onto the same identifier and have their events incorrectly correlated.

Solution: Encode the 128-bit identifiers as UUID strings before sending. A UUID carries the same 128 bits losslessly while remaining mathematically equivalent to the original integer (uuid.UUID(int=value).int == value). This preserves full fidelity while keeping new events linkable to pre-migration history.

Changes

1. reflex/utils/telemetry.py

   • Added _encode_distinct_id() function to convert 128-bit integers to canonical UUID hex strings
   • Updated _Properties TypedDict to declare distinct_id and distinct_app_id as str instead of int
   • Modified _get_event_defaults() to encode both identifiers via _encode_distinct_id()

2. reflex/utils/prerequisites.py

   • Changed ensure_reflex_installation_id() to generate uuid.uuid4().int instead of random.getrandbits(128)
   • Added comment explaining that the integer form is stored for backward compatibility with older Reflex versions

3. reflex/utils/frontend_skeleton.py

   • Changed init_reflex_json() to generate uuid.uuid4().int instead of random.getrandbits(128)
   • Added comment explaining the encoding strategy

4. tests/units/test_telemetry.py

   • Updated event_defaults fixture to expect UUID strings instead of integers
   • Added test_encode_distinct_id_round_trips_losslessly() to verify 128-bit precision is preserved
   • Added test_encode_distinct_id_handles_uuid4_int_form() to verify uuid4 round-tripping
   • Added test_encode_distinct_id_pads_small_values() to verify small integers are zero-padded correctly
   • Added stub_event_default_sources fixture to mock slow/host-specific telemetry inputs
   • Added test_get_event_defaults_encodes_ids_as_uuid_strings() regression test
   • Added test_get_event_defaults_omits_distinct_app_id_without_project_hash() to verify conditional behavior
   • Added test_get_event_defaults_returns_none_without_installation_id() to verify existing contract

Test Plan

All new unit tests pass and cover:

• Lossless round-trip encoding/decoding of 128-bit identifiers
• UUID4 integer form handling
• Zero-padding of small values
• Integration with _get_event_defaults()
• Conditional omission of distinct_app_id when project hash is unavailable
• Preservation of existing behavior when installation ID is missing

Existing telemetry tests updated to expect UUID string format.

https://claude.ai/code/session_0162Wc1GmkskbgCRs7fjg9Cy

[greptile-apps]

Greptile Summary

Fixes a telemetry bug where 128-bit installation and project identifiers were being sent as raw JSON integers, which PostHog silently coerces to floats, discarding all but ~16 significant digits and collapsing distinct installs/apps onto the same truncated ID.

• Adds _encode_distinct_id() in telemetry.py to encode stored 128-bit integers as canonical UUID hex strings before they leave the process, updating _Properties and _get_event_defaults() accordingly.
• Replaces random.getrandbits(128) with uuid.uuid4().int in both prerequisites.py and frontend_skeleton.py; the integer form is still persisted to disk for backward compatibility with older Reflex versions.
• Six new unit tests cover lossless round-trip encoding, zero-padding of small values, and regression coverage of _get_event_defaults() emitting string IDs.

Confidence Score: 5/5

Safe to merge — the change is narrowly scoped to telemetry ID encoding, all error paths are wrapped by the existing suppress(Exception) in _send, and new tests verify the round-trip fidelity.

The encoding logic is minimal (str(uuid.UUID(int=value))), the TypedDict update is consistent, generated IDs remain backward-compatible on disk, and the test suite directly covers the regression scenario. No application logic outside telemetry is affected.

No files require special attention.

Important Files Changed

Filename	Overview
reflex/utils/telemetry.py	Adds _encode_distinct_id() to convert 128-bit ints to UUID strings; updates _Properties TypedDict and _get_event_defaults() to use it.
reflex/utils/prerequisites.py	Swaps random.getrandbits(128) for uuid.uuid4().int when generating a new installation ID, preserving integer storage for backward compatibility.
reflex/utils/frontend_skeleton.py	Same random.getrandbits(128) → uuid.uuid4().int change for project hash generation; comment explains encoding strategy.
tests/units/test_telemetry.py	Updates existing fixture to expect UUID strings; adds six new targeted tests covering round-trip fidelity, zero-padding, and _get_event_defaults() integration.

_{Reviews (1): Last reviewed commit: "fix(telemetry): send distinct ids as UUI..." | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will improve performance by 3.68%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
✅ 25 untouched benchmarks
⏩ 8 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_var_access[mutable_dataclass_list]	218.2 ms	210.4 ms	+3.68%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing claude/compassionate-planck-wjwQy (097f506) with main (00fdeaf)}

Footnotes

1. 8 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩