Is it possible to use Python 3.8 ?

gnaldi62

Hi all,

sorry, but cannot find an answer to this. First at all, we know that Python 3.8. is not supported, nevertheless

we'd like to know if it would be possible to use it anyway, and if it is possible to create a code env pointing to

that installation . We've tried with "Managed by DSS" specifying the full executable path but

it gives an error related to the Jupyter notebook support. Please, advice. Many thanks in advance. Rgds.

Giuseppe

HarizoR

Hi Giuseppe,

As you correctly stated, as of today Python 3.8 is not supported by DSS. In practice, it means that you cannot use Python 3.8 with DSS, even through a code environment.

Supporting Python 3.8 is already in our development backlog so I will add your use-case to the feature request.

Best,

Harizo

gnaldi62

Thanks for your quick reply. Giuseppe

MRvLuijpen

Hello Dataiku,

Is there any update on this subject Using Python 3.8?

Thanks in advance

Marc Robert

HarizoR

Hi,

No change since my last message, the support of Python 3.8 is still being worked on for a future version.

Best,

Harizo

tgb417

@HarizoR
,

With the release of DSS V11 have we gotten to the point that we can support V3.8+ as the base python that DSS is using? Or do we still need to use Python V3.7?

Thanks for the help.

Turribeach

Even in v11 the base DSS Python environment continues to be v2.7. Since v10.0.4 (March 7th, 2022) Dataiku supports Python 3.8, Python 3.9 and Python 3.10 (experimental) but only for code environments. On my latest Dataiku v11 environment I have Python v2.7, v3.6, v3.7, v3.8 and v3.9 all installed and living peacefully amond each other. And I can create code environments on any Python version I want. So the fact that DSS uses Python v2.7 is somehow a problem since it is unsupported but it doesn't really affect our flows as we can choose to run our flows using new Python versions.

tgb417

@Turribeach

it good to hear that you can get all of those different pythons working together. I’m still working on getting this all to work together. I’ve got the basic dss working. But I’m running into some problems getting python 3.6 working.

I’m wondering on what OS you have made this work, one of the Linux distorts I’d guess or are you over on a Macintosh?

Are you using a python version manager at the os level, something like pyenv, miniconda, anaconda, conda/mini forge or something else.

tgb417

In my tests today, it seem like 3.8 is still not ready as a base python version. It still looks like we need Either Python 3.6.x or 3.7.x

Turribeach

Hi Tom, I am on RHEL 7.9. I suggest you start a new thread and post all the details of what issue are you seeing, how are you installing the different Python versions, what your OS is, etc. It's not clear at all what you mean by "it seem like 3.8 is still not ready as a base python version".

HarizoR

Hi tgb417,

As of Dataiku 11, the supported base environment versions for Python 3 are still 3.6 and 3.7. That being said, for code environments you still can leverage Python 3 versions up to 3.10.

If you have use-cases for which you absolutely require a version of Python higher than 3.7 for the base environment, don't hesitate to provide further explanations.

Best,

Harizo

tgb417

@HarizoR
,

My use case is related to Apple Silicon. Pythons 3.8 and above are compiled for arm processors. Thus providing better performance than earlier versions that are not. (And yes, I know that I’m coloring outside of the lines, playing with a configuration like this.) Looking forward to future support of Apple silicon in future versions of DSS.

HarizoR

Hi,

In this case, as stated in https://community.dataiku.com/t5/Setup-Configuration/DSS-installed-on-Mac-M1/m-p/27513/highlight/true#M2505 you can install DSS directly using the Launcher. This installation mode is compatible with Apple M1 computers.

Best,

Harizo