community
cancel
Showing results for 
Search instead for 
Did you mean: 

Alteryx Designer Ideas

Share your Designer product ideas - we're listening!

Jupyter Notebooks Should Clear Caches Correctly

It has become clear that the Jupyter Notebook integration caches code and does not appropriately clear when there are changes made - resulting in "saved" workflows that do not contain updated code. This happens when two people are using a "shared" workflow (emailed back and forth or from a shared drive) if one person does not completely shut down out of Designer Desktop if they had previously had the workflow open at any point. This has been confirmed by Alteryx Support and is not just my hunch. 

 

This also happens sometimes with a single user - where the Jupyter Notebook save button has been pressed multiple times and the workflow has been saved, but the changes do not make it to the file. 

 

The integration is a step in the right direction for sure and is great to use - but my idea is that the cache should be attached to the workflows, not the entire session of Designer. Not knowing if changes were actually saved, and discovering that some were not is extremely frustrating. 

6 Comments

It is critical that this behavior is fixed or the tool is not functional -- actually dangerous -- with multiple people working together editing workflows.

Alteryx
Alteryx

Hi alentz, 

 

Thanks for your feedback. We will look into this issue. 

 

-Tyler

Alteryx
Alteryx

Dear Alentz,

 

We began investigating this issue and the engineering team had a few comments and questions.

 

Currently, Alteryx does not support multiple users editing the same workflow files concurrently. For example, if you open and edit a workflow file from a shared location, no other user should be editing that same file. When opening a workflow from email, we recommend you save a copy to your local system and then edit that copy. Some enterprises utilize shared location for managing email attachments which could explain the behavior you’re describing.

 

We have a couple of questions for you to help us better understand your use case:

 

  1. When you say, “the changes do not make it to the file”, do you mean the workflow xml or the Jupyter notebook file? What are the changes missing? Is it the entire notebook contents, or all changes since your last save?
  2. When sharing a workflow via email, do you ever make changes to the file after attaching it?

 

-Sonia 

Atom

Sonia, 

 

We are not "concurrently" editing workflows in that we do not have the same workflow open at the same time from the same location. We are simply opening the same file from the same location at different times. I haven't checked the XML or the underlying Jupyter notebook, but when you reopen the file and look at the notebook, the code changes have not saved and an older version appears. Is there a way to check the underlying Jupyter notebooks from outside of Alteryx? 

 

The only way we have been able to override this behavior is to completely close out of Alteryx Designer after every save or before opening a workflow that was updated. This would indicate to me that the notebooks are cached per instance of Alteryx Designer, not by workflow, since simply closing the workflow does not override this behavior. 

Moderator
Moderator

Hi @alentz,

 

Thank you for your comments. Could you please send an email to support@alteryx.com and join a link to current thread so we could investigate/escalate?

 

I would also need you to specify the following:

 

- what is your version of Alteryx Designer please?

- when issue occurs with multiple users, do they use Designer on same machine?

- what is the value for default temporary directory (as per User Settings, Edit User Settings, Defaults tab)?

 

 

Best,

 

Paul Noirel

Sr Customer Support Engineer

Atom

Hi Paul, 

 

Respectfully, I will not open another support ticket over this issue. I have already worked with support, who confirmed this functionality and suggested I post my "idea" on the forum (so I did). 

 

I have done my best to replicate this scenario and other scenarios where the notebooks load cached code that they should not. Please see my attached video. 

 

Regards, 

A