Timeline for sempy_labs -> sempy

[OldDogNewTricks000]

Is there any visibility to when functions will migrate from sempy_labs -> sempy?

I love the functionality of sempy_labs, however I don't love the additional overhead of spinning up an environment with it installed instead of using a default warm environment or pip installing if running in an interactive way. That is why I was wondering what the roadmap looks like to move functions from sempy_labs -> sempy.

[m-kovalsky]

Semantic Link Labs will always have capabilities which don't exist in semantic link so you will have to pip install semantic link labs. That being said, just because you don't have to pip install semantic link (since it is pre-installed in the Fabric runtime), does not mean that you have the latest version. If you want the latest version of semantic link you will likely have to pip install that as well. Anyways, pip install for labs takes about 6 seconds in a pure python notebook so I don't see that as a limitation. In an ideal world all would be in a single library but the development speed of labs can naturally be much faster than semantic link - hence our current scenario.

[OldDogNewTricks000]

I understand what you are saying. I just don't like the overhead of using an environment because we cannot use a warm pool, so that is closer to about a 2 min session start time vs about 10 sec for a warm default pool. That is why I was asking about the roadmap at the function level from semantic-link-labs -> semantic-link (and runtime version included).

[m-kovalsky]

Honestly I don’t recommend using environments with custom libraries. It’s much faster to just pip install. The process for functions to move to semantic link is out of my control. First up is SP auth support and wrapper functions for pbi & fabric APIs. However I don’t have a timeline for that.

[OldDogNewTricks000]

I have not been able to use pip install w/ scheduled notebooks; I receive an error. Is there a way to use pip install from a scheduled notebook?

[m-kovalsky]

Ah for a scheduled notebook then yes a custom environment is easier. You may be able to do !pip install in that scenario instead of %pip install but it’s probably not worth it. Honestly I have never executed a scheduled notebook.