airbytehq / airbytehq/PyAirbyte

🐛 Bug: Warnings looks like errors: `Could not determine airbyte type from JSON schema`

Ouverte
#213 2 commentaires 1 réaction 0 personnes assignées Voir sur GitHub
Langage dominant
Python
Étoiles
344
Forks
77
Merge moyen
1 j 11 h
PR mergées (30 j)
35

Description

We've received this a few times now - when PyAirbyte cannot determine a type with confidence, it prints a warning message but users often are led to believe this is fatal. We should improve the messaging here so it is more clear that this is a warning - or we should remove the messaging altogether.

Raised (most recently) here:

- https://github.com/airbytehq/PyAirbyte/issues/212

Background:

Whenever PyAirbyte determines a column type, it uses a heuristic to try to determine the "best" column type given the input data type. However, PyAirbyte always applies a "failsafe" data type when none can be determined otherwise. This is normally a string, JSON, or Variant column type, and in most cases the sync will succeed regardless - with no real loss of fidelity.

Example cases:

A classic example of this is a data type that is defined as `anyOf(string, object)`. Since databases generally don't support a column type that would equally support either a string or an object (aka dictionary/struct), the cache will determine the best-available failover type, which would probably be `VARCHAR` - meaning object-type inputs would be converted to string. This works fine in practice; while we might lose some _"data type fidelity"_ we generally have zero loss of _"data fidelity"_.

Clarified messaging:

The right thing to do here (apart from adding more advanced heuristics) would be to improve messaging so that the user knows that this is _not_ a failure condition, and that we do have a valid typing fail-safe that we are applying for this condition. Alternatively, we could eliminate this messaging entirely.

cc @marcosmarxm , @bindipankhudi

Guide de contribution

Ouvrir le guide de contribution

Évaluation

Cette issue n'a pas encore été évaluée.

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.