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
This came up in a comment on my recent PR: #739 (comment)
Is your feature request related to a problem?
The aws_lambda instrumentation does not need to have any dependency packages installed for it to instrument AWS Lambda, but obviously is only applicable in the AWS Lambda environment.
This means with no dependencies, it is on the same design as instrumentations of stdlib packages. However, stdlib packages should always be instrumented because they're packages are available in the standard distribution of python.
Describe the solution you'd like
As suggested by @owais, we could add a opentelemetry.instrumentation.symbols.STDLIB symbol to distinguish them.
Describe alternatives you've considered
Which alternative solutions or features have you considered?
Additional context
In a quick search of the specifications I didn't find mention of this. Should I be discussing this there? It seems like something other SIGs would run into.
The text was updated successfully, but these errors were encountered:
One more thing to consider is that right now there is no way to indicate that an instrumenation should be installed if just one out of all mentioned libraries is found. For example, opentelemetry-instrumentation-psycopg works with both psycopg2 and psycopg2-binary but we only list one as a target today. skit-learn also has an alias called sklearn.
This is not a related problem but if we are talking about improving how this metadata is declared, it'd probably be helpful to keep all use cases in mind.
This came up in a comment on my recent PR: #739 (comment)
Is your feature request related to a problem?
The
aws_lambda
instrumentation does not need to have any dependency packages installed for it to instrument AWS Lambda, but obviously is only applicable in the AWS Lambda environment.This means with no dependencies, it is on the same design as instrumentations of stdlib packages. However, stdlib packages should always be instrumented because they're packages are available in the standard distribution of python.
Describe the solution you'd like
As suggested by @owais, we could add a
opentelemetry.instrumentation.symbols.STDLIB
symbol to distinguish them.Describe alternatives you've considered
Which alternative solutions or features have you considered?
Additional context
In a quick search of the specifications I didn't find mention of this. Should I be discussing this there? It seems like something other SIGs would run into.
The text was updated successfully, but these errors were encountered: