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
In #994 the SDK was changed such that the WSGI integration and all frameworks that inherit from it have "request-mode" sessions on by default.
However, despite the default auto_session_tracking=True init option, session tracking is not enabled for anything outside of the WSGI realm.
Missing:
Automatic "application-mode" session for non-Web-servers applications
Automatic session for non-WSGI integrations (e.g., Django and ASGI)
These are also missing but probably should actually disable sessions altogether (and have accompanying documentation):
Queue consumers (e.g., Celery and RQ) ⇒ it is unclear doing 1 session per queue item is justified, there is no industry precedence on this to set user expectations
Serverless (e.g., AWS Lambda and Google Cloud Functions) ⇒ considering that the current implementation requires a separate request to send a session update, this would require a minimum of 2 requests per cloud function invocation (1 for transaction and 1 for session) which would be highly inefficient.
We need to document what auto_session_tracking enables, clarifying what the behavior is.
The text was updated successfully, but these errors were encountered:
This issue has gone three weeks without activity. In another week, I will close it.
But! If you comment or otherwise update it, I will reset the clock, and if you label it Status: Backlog or Status: In Progress, I will leave it alone ... forever!
"A weed is but an unloved flower." ― Ella Wheeler Wilcox 🥀
In #994 the SDK was changed such that the WSGI integration and all frameworks that inherit from it have "request-mode" sessions on by default.
However, despite the default
auto_session_tracking=True
init option, session tracking is not enabled for anything outside of the WSGI realm.Missing:
These are also missing but probably should actually disable sessions altogether (and have accompanying documentation):
Queue consumers (e.g., Celery and RQ) ⇒ it is unclear doing 1 session per queue item is justified, there is no industry precedence on this to set user expectations
Serverless (e.g., AWS Lambda and Google Cloud Functions) ⇒ considering that the current implementation requires a separate request to send a session update, this would require a minimum of 2 requests per cloud function invocation (1 for transaction and 1 for session) which would be highly inefficient.
We need to document what
auto_session_tracking
enables, clarifying what the behavior is.The text was updated successfully, but these errors were encountered: