Semantic link vs. Semantic link labs vs. Sempy vs. Sempy labs

[frithjof_v]

Hi, I'm trying to understand how these relate to one another.

Is this how it works:

- Semantic link is the foundational library

- Sempy is a package within Semantic link 

- Semantic link labs is an extension built upon Semantic link

- Sempy labs is a package within Semantic link labs

I'm a Power BI guy and my understanding of "libraries" and "packages" is limited. 

Could you explain how Semantic link, Sempy, Semantic link labs, and Sempy labs relate to each other from a beginner's perspective? Also, when should I use each name?

Thanks!

[frithjof_v]

Thanks @v-shex-msft,

Is semantic link labs only in experimental status? (Preview status)? 

Is it not considered stable?

I'm curious about this 

[v-shex-msft]

HI @frithjof_v,

It means this is a lab version of semantic link, so they semantic link features should also works smoothly.

For the experimental part of early access features which I mentioned, it is hard fully ensure they always works in stable.  (Also, I suppose the experimental version popular features, they may be released to the original package)

Regards,

Xiaoxin Sheng

[v-shex-msft]

Hi @frithjof_v,

In fact, sempy is a package that include semantic link features, it also provides data operations while preserving the semantic information from models.
For the labs versions, normally they will provide more features than original versions.
So if you only want to connect semantic models with data science tools in Microsoft Fabric, you can choose semantic link. If you also need to apply some data operations with this model, you can choose sempy.
What is semantic link? - Microsoft Fabric | Microsoft Learn
If you not only want to work with semantic model, but also need to manage with some tabular object models, you can consider to use semantic link labs.

GitHub - microsoft/semantic-link-labs: Early access to new features for Microsoft Fabric's Semantic ...
If you also want to experience some experimental features of sempy, you can consider to using labs version.

Regards,
Xiaoxin Sheng