Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SPARK-51073][SQL] Remove Unstable from SparkSessionExtensionsProvider trait #49780

Closed
wants to merge 1 commit into from

Conversation

dongjoon-hyun
Copy link
Member

@dongjoon-hyun dongjoon-hyun commented Feb 3, 2025

What changes were proposed in this pull request?

This PR aims to remove Unstable from SparkSessionExtensionsProvider trait from Apache Spark 4.1.0.

@DeveloperApi
@Unstable
@Since("3.2.0")
trait SparkSessionExtensionsProvider extends Function1[SparkSessionExtensions, Unit]

Why are the changes needed?

Since Apache Spark 3.2.0, we provide SparkSessionExtensionsProvider trait without any change to allow third-party developers to provide a resource file that contains default extensions.

Apache Iceberg has been using this since 0.11.0.

Although SparkSessionExtensions is an unstable developer API, SparkSessionExtensionsProvider trait itself and ServiceLoader framework is stable and it's unlikely for Apache Spark to change this trait.

@DeveloperApi
@Experimental
@Unstable
class SparkSessionExtensions {

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Pass the CIs.

Was this patch authored or co-authored using generative AI tooling?

No.

@github-actions github-actions bot added the SQL label Feb 3, 2025
@dongjoon-hyun
Copy link
Member Author

dongjoon-hyun commented Feb 3, 2025

Hi, @yaooqinn . WDYT about this?

@dongjoon-hyun dongjoon-hyun changed the title [SPARK-51073][SQL] Remove Unstable from SparkSessionExtensionsProvider [SPARK-51073][SQL] Remove Unstable from SparkSessionExtensionsProvider trait Feb 3, 2025
@@ -76,7 +74,6 @@ import org.apache.spark.annotation.{DeveloperApi, Since, Unstable}
* @since 3.2.0
*/
@DeveloperApi
@Unstable
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh actually, I would prefer to have this as unstable because it can touch stuff like optimizer and analyzer rules that are internal API..

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I understand we won't likely remove SparkSessionExtensions, and this is service loader for SparkSessionExtensions (as mentioned in PR description). But I would prefer to avoid promoting this.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you elaborate a little more about how SparkSessionExtensionsProvider are affected with a change of SparkSessionExtensions? IIUC, we are not going to remove or rename class SparkSessionExtensions .

trait SparkSessionExtensionsProvider extends Function1[SparkSessionExtensions, Unit] 

ServiceLoader is only related to the relation between the text resource file and class name, SparkSessionExtensions, isn't it?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe, are you considering removing this ServiceLoader pattern, @HyukjinKwon ?

Copy link
Member

@HyukjinKwon HyukjinKwon Feb 4, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I mean, nothing will be affected by this ... I just worry that more people may SparkSessionExtensions itself...

TBH, I personally don't like this SparkSessionExtensions in general because it opens too much catalyst API usage, and touching rules.. and I was thinking about having them all as Unstable for now..

I actually don't like all those API ... such as df.queryExecution, etc. for the same reason.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am fine if others are okay - I don't have a strong opinion.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, I understand your concerns fully. I also understand why this is in this status. However, Apache Spark depends on this since v2.2.0 (SPARK-18127) because this is the only way for certain features.

The goal of this PR is only this ServiceLoader pattern is applicable in the stable contract for the production usage (SRE or dev team) in Spark 4.1.0.

If many people are reluctant, I'm also going to drop this~ So, let's see.

Copy link
Member

@yaooqinn yaooqinn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, the trait itself can be considered stable.

@dongjoon-hyun
Copy link
Member Author

dongjoon-hyun commented Feb 10, 2025

Thank you, @yaooqinn and @HyukjinKwon .

Merged to master for Apache Spark 4.1.0.

@dongjoon-hyun dongjoon-hyun deleted the SPARK-51073 branch February 10, 2025 19:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants