✈ Pre-Flight checks
🐞 Describe the bug
Summary
Three fields in the [telemetry] section of pyRevit_config.ini accumulate escape characters on every save, doubling in size each cycle. After ~22 reload cycles they reach approximately 8 MB each, bloating the config file to ~25 MB. This dramatically slows PyRevitConfig.Load() and indirectly slows pyRevit reload to several minutes (vs ~40s for a clean config).
Affected fields
telemetry_file_dir
telemetry_server_url
apptelemetry_server_url
Other telemetry fields in the same section (utc_timestamps, active, active_app, apptelemetry_event_flags, include_hooks) are not affected. Other string fields elsewhere in the config (extension paths, clone paths, tab colors, last window geometry) are also not affected — the bloat is specific to these three.
⌨ Error/Debug Message
After many cycles, the value looks like:
telemetry_server_url = "\"\\\"\\\\\\\"\\\\\\\\\\\\\\\"\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\"..."
♻️ To Reproduce
- Open %APPDATA%\pyRevit\pyRevit_config.ini
- Note the current value of telemetry_server_url (or set it to anything containing characters that need JSON-escaping, such as a URL with /)
- Open Revit with pyRevit attached → reload pyRevit → close Revit
- Inspect the value — additional escape sequences will have been added to the stored representation
- Repeat steps 3–4 several times — the value grows exponentially
⏲️ Expected behavior
No response
🖥️ Hardware and Software Setup (please complete the following information)
pyRevit: 6.4.0.26098+2214 (current develop branch tip)
Revit: 2024.3.4 / 2026 FCS / 2027
OS: Windows 10 build 26200
CLI version: matches develop tip
Additional context
The escape pattern doubles in depth on each save: " → \" → \\\" → \\\\\\\" etc. This indicates the writer is re-escaping previously-escaped characters as if they were literal content of the value.
Impact
Even with telemetry fully disabled (active = false and active_app = false, which is the default), the bloated values still get parsed on every PyRevitConfig.Load(). With ~25 MB of escape-heavy text to process, this:
Adds 30–60s to first config read (cold cache)
Likely triggers pathological backtracking in whatever regex/parser handles these specific values
Cascades into reload performance — observed ~3 min reload vs ~40s baseline, where the only difference between baseline and bloated state was the size of these three fields
Reload time was fully restored to ~40s after blanking the three fields manually, confirming these values are the sole cause.
Workaround
Edit the config file directly while Revit is closed, set the three fields to empty strings:
initelemetry_file_dir = ""
telemetry_server_url = ""
apptelemetry_server_url = ""
Since telemetry is disabled by default, these values serve no functional purpose for most users. The bloat will resume on subsequent saves until the underlying serialization bug is fixed.
Suggested investigation area
The escape mismatch likely lives in the INI serializer used for these specific fields — possibly in pyRevitLabs.Common or wherever string fields with embedded special characters get persisted. The asymmetry is between:
Read path: does not unescape the previously-written escape characters (treats them as literal value content)
Write path: does JSON-encode the value before writing (adding another layer of escapes on top of what was read)
The result is that the value grows by one escape level per save cycle.
The fact that other string fields don't bloat suggests these three go through a different code path — perhaps a JSON-typed field that was unintentionally double-wrapped, or a string-typed field that was escape-encoded once at creation and the writer doesn't recognize the existing encoding.
✈ Pre-Flight checks
🐞 Describe the bug
Summary
Three fields in the [telemetry] section of pyRevit_config.ini accumulate escape characters on every save, doubling in size each cycle. After ~22 reload cycles they reach approximately 8 MB each, bloating the config file to ~25 MB. This dramatically slows PyRevitConfig.Load() and indirectly slows pyRevit reload to several minutes (vs ~40s for a clean config).
Affected fields
telemetry_file_dir
telemetry_server_url
apptelemetry_server_url
Other telemetry fields in the same section (utc_timestamps, active, active_app, apptelemetry_event_flags, include_hooks) are not affected. Other string fields elsewhere in the config (extension paths, clone paths, tab colors, last window geometry) are also not affected — the bloat is specific to these three.
⌨ Error/Debug Message
After many cycles, the value looks like: telemetry_server_url = "\"\\\"\\\\\\\"\\\\\\\\\\\\\\\"\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\"..."♻️ To Reproduce
⏲️ Expected behavior
No response
🖥️ Hardware and Software Setup (please complete the following information)
Additional context
The escape pattern doubles in depth on each save: " → \" → \\\" → \\\\\\\" etc. This indicates the writer is re-escaping previously-escaped characters as if they were literal content of the value.
Impact
Even with telemetry fully disabled (active = false and active_app = false, which is the default), the bloated values still get parsed on every PyRevitConfig.Load(). With ~25 MB of escape-heavy text to process, this:
Adds 30–60s to first config read (cold cache)
Likely triggers pathological backtracking in whatever regex/parser handles these specific values
Cascades into reload performance — observed ~3 min reload vs ~40s baseline, where the only difference between baseline and bloated state was the size of these three fields
Reload time was fully restored to ~40s after blanking the three fields manually, confirming these values are the sole cause.
Workaround
Edit the config file directly while Revit is closed, set the three fields to empty strings:
initelemetry_file_dir = ""
telemetry_server_url = ""
apptelemetry_server_url = ""
Since telemetry is disabled by default, these values serve no functional purpose for most users. The bloat will resume on subsequent saves until the underlying serialization bug is fixed.
Suggested investigation area
The escape mismatch likely lives in the INI serializer used for these specific fields — possibly in pyRevitLabs.Common or wherever string fields with embedded special characters get persisted. The asymmetry is between:
Read path: does not unescape the previously-written escape characters (treats them as literal value content)
Write path: does JSON-encode the value before writing (adding another layer of escapes on top of what was read)
The result is that the value grows by one escape level per save cycle.
The fact that other string fields don't bloat suggests these three go through a different code path — perhaps a JSON-typed field that was unintentionally double-wrapped, or a string-typed field that was escape-encoded once at creation and the writer doesn't recognize the existing encoding.