Sign up to take part
Registered users can ask their own questions, contribute to discussions, and be part of the Community!
I'm experiencing a regression on Dataiku between V9.0.1 and V12.1.0
Indeed, we were on the V9.0.1. I used a Join recipe with a Date join key named "D_ARRETE" parsed correctly on Dataiku, I applied D_ARRETE=D_ARRETE in the Join part in the recipe. This recipe join worked well and returned good results with no missing value.
After upgrading Dataiku to version 12.1.0, I noticed this recipe runs fine as before but returned columns without values (empty fields), after investigating this behavior, I understood that following the v12 upgrade, the Join recipe no longer takes into account the D_ARRETE=D_ARRETE rule that I defined, the recipe no longer finds a match between the D_ARRETE values between the two tables. This explains the empty fields returned.
So I had to add pre-computed columns D_ARRETE_TECH and
D_ARRETE_max_TECH By putting this SQL query :
CAST(D_ARRETE AS DATE FORMAT 'YYYY-MM-DD')
This fixed my problem, the Join recipe returned the correct values with no blank fields.
This problem has been encountered by other users of the teams I support.
This is problematic for the Dataiku users that I am currently supporting as part of my mission. This is a glaring regression from the V9.
Thank you for looking at the topic. I remain available for any further information.
Operating system used: Windows 11
Sounds to me like your join type or the data type of your join columns may have changed in v12 for some reason. 12.1.0 is quite old, did you test with 12.3.2 to confirm the issue is still there in the latest v12 release?
I am a Dataiku user and I cannot upgrade to a newer version, as our administrative team does not plan any further updates.
Additionally, I would like to clarify that I have not changed my dates or any part of my workflow, which has been the same since version 9 I noticed this issue after updating to version 12.1.0