Auto-instrumentation, getting-started guide, and end user experience

[srikanthccv]

In my experience of interacting with the community, I have noticed that it can be overwhelming for (first-time) users to learn about concepts like provider, propagator, exporter, OTLP, etc... And a lot of them want to instrument their applications with minimal changes. Auto instrumentation eliminates a lot of these pain points and I believe we have a reasonable default configuration. Starting this discussion in an effort to make using opentelemetry python a seamless effort. Please share your thoughts on what can be done to improve this experience? Some things we can address in near future. 1) Why doesn't the auto instrumentation package depend on and install the distro package? People run into this very often #1785, #1904. 2) We don't have any quickstart that can help people to get started without learning all the jargon.

[lzchen]

It was even slightly difficult for me to get started with auto-instrumentation so +1 for the quickstart, especially with a very simple hello_world example (not flask, etc).

I believe we chose not to depend on the default distro because we wanted the user to have the option of choosing which distro to use (since installing one will automatically find and use the first one when auto-instrumenting). This might be solved with this.

Although in the meantime, we should probably make it very explicit that you need SOME distro installed.