Remove pinned babel==2.9.1 reference.
Version
4.17.0 4.16.2 4.16.1 4.16.0 4.15.1 4.15.0 4.14.8
Describe the bug
Every single version for quite some time has a PINNED dependency on babel==2.9.1. Multiple packages already require babel>=2.10.
This has been the case since October 2021 (https://github.com/microsoft/botbuilder-python/pull/1826) and still pinned to that babel version. Bad practice.
This is creating a detrimental cascading effect on more complex projects. Having teams-ai depend on this botbuilder library, makes things more relevant.
Right now, one has to resort to hack-install babel 2.1x and it works, but pip beggining with version 25 (I think) will not willingly allow that. Errs out.
Deploying to an automated container becomes then a challenge, because well, it's automated and, automatically fails.
To Reproduce
Steps to reproduce the behavior:
- Add teams-ai to a python project -or-
- Add botbuilder-python to a python project.
- In vscode create a Jupyter notebook.
- Set the environment to run on a 3.12 kernel.
- Create a cell on that notebook... i.e.
a=37 - Run the cell. --> Runs ok.
- Install jupyterlab (
pip install jupyterlab) for some data science graphing for example. --> Jupyterlab installer replaces babel 2.9.1 with 2.1x
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
botbuilder-dialogs 4.17.0 requires babel==2.9.1, but you have babel 2.17.0 which is incompatible.
Additional notes
Has been tested and appears to work ok having babel>=2.10 installed. But, has to be a forced, manual installation. This is just an example not very common but a library should not be pinned to another - for years.
Expected behavior
Change https://github.com/microsoft/botbuilder-python/blob/main/libraries/botbuilder-dialogs/setup.py, line 15 from "babel==2.9.1" to "babel>=2.9.1" or somehting around those lines.
Would not hurt to remove the pinned reference to emoji==1.7.0 which is extremly old. By far.
@axelsrz - Can you take a look at this?