You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Oct 23, 2023. It is now read-only.
We want to avoid issues like #295 as much as possible, so we're going to start automatically exercising the most memory-leak-prone parts of raven as a sanity/regression test for memory usage. The plan is to build a harness that lets us just run a bunch of calls through whatever thing (contexts, instrumentation, etc) and then follow memory with memwatch and fail a test case if numbers don't look right. A loose skeleton of this is already done from tracking down the memory leak in #295.
Things to do this check for:
Having lots of different ephemeral contexts
Lots of manual captureBreadcrumbing
each autoBreadcrumb instrumentation
console, http, pg
any added in future
The goal is that if these tests had been in place, the change that caused #295 would not have passed CI. This will also increase confidence in future additional instrumentation.
The text was updated successfully, but these errors were encountered:
We built these memory tests in #300 except for the "automated" part. Did not automate because it's hard to programmatically know whether the numbers the tests produce are acceptable; it's easier and good enough IMO to, before cutting any release, manually run the memory tests and take a look over the numbers.
We want to avoid issues like #295 as much as possible, so we're going to start automatically exercising the most memory-leak-prone parts of raven as a sanity/regression test for memory usage. The plan is to build a harness that lets us just run a bunch of calls through whatever thing (contexts, instrumentation, etc) and then follow memory with memwatch and fail a test case if numbers don't look right. A loose skeleton of this is already done from tracking down the memory leak in #295.
Things to do this check for:
captureBreadcrumb
ingThe goal is that if these tests had been in place, the change that caused #295 would not have passed CI. This will also increase confidence in future additional instrumentation.
The text was updated successfully, but these errors were encountered: