aws-sdk-core >= 3.192.0 may be not compatible with AWS SQS APIs in XML format

[yokonao]

Describe the bug

Our application uses aws-sdk-core 3.170.0 and aws-sdk-sqs 1.46.0.
Recently, we try to update aws-sdk-core to 3.196.1. However, we canceled the update because some features using AWS SQS were degraded.

As a result of investigation, it was found that the result of SQS ReceiveMessage was unintentionally nil.

I did some more detailed debugging and found the following:

• Our application uses AWS SQS APIs in XML format.
  • When aws-sdk-sqs >= 1.66.0, APIs in JSON format are selected. (commit)
• When aws-sdk-core is 3.191.6, we can receive the message. However, when aws-sdk-core is 3.192.0 we cannot.
• AWS XML Parser changes in 3.192.0. (Pull Request)
• When the changes of the above PR is reverted, we can receive the message.

Expected Behavior

We can receive the message in AWS SQS when aws-sdk-core >= 3.192.0 and aws-sdk-sqs < 1.66.0.

Current Behavior

We CANNOT receive the message in AWS SQS when aws-sdk-core >= 3.192.0 and aws-sdk-sqs < 1.66.0.

Reproduction Steps

I prepared a sample repository for reproducing this issue.
https://github.com/yokonao/aws-sdk-ruby-examples?tab=readme-ov-file#sqs-xml

Possible Solution

No response

Additional Information/Context

No response

Gem name ('aws-sdk', 'aws-sdk-resources' or service gems like 'aws-sdk-s3') and its version

aws-sdk-core and aws-sdk-sqs

Environment details (Version of Ruby, OS environment)

Ruby 3.1.5, macOS Ventura

[mullermp]

We're aware of the sweeping protocol changes that affected xml in some way. This was an organizational effort to align with expected behavior. There is not much we can do about it now. My question is, why are you trying to upgrade core and not the sqs gem? If you were to upgrade both, you shouldn't see any issue.

[yokonao]

@mullermp
Thanky for your reply!

I understand updating both aws-sdk-sqs and aws-sdk-core is the best solution. We are working towards that.
However, in order to update aws-sdk-sqs, several other prerequisites must be met, so it may take some time. (For example, localstack v2 does not support AWS SQS APIs in JSON format.)
And we were motivated to apply patches for aws-sdk-core iteration issues sooner. (#3015, #3021)

Additionally, our application is so huge that updating all AWS SDK gems at the same time is difficult.

These are why we have tried to update aws-sdk-core only.

This was an organizational effort to align with expected behavior. There is not much we can do about it now.

Does this mean that the backward compatibility break reported in this issue is intentional and cannot be fixed?

[mullermp]

I understand. I am very sorry about this. Once localstack is updated to support SQS JSON, you should be able to upgrade SQS and core - no need to upgrade your other SDK gems unless you want to (you should)

Have you created an issue or feature request? I can also poke the author to state that this is now expected, and we can possibly help with code.

It was not intended to break, however, we've only heard of 2 other cases of SQS related breakage. Given that was minor in that only a few customers were impacted, we decided to commit with the overall changes. Those other customers were able to upgrade easily.

For now, you can possibly get away with upgrading aws-sdk-core while also doing a monkey patch. I'm not exactly sure what affected method you would need to patch, but my guess would be xml_name in Aws::Xml::Parser::Frame. You can try monkey patching the method to be what it used to be in https://github.com/aws/aws-sdk-ruby/pull/2988/files#diff-576fcdf895f808f67ab8e4d1b4993ae6598414e23be5fd8c58063aa8858212fd.

[yokonao]

Thank you!

Our basic policy at the moment is as follows.

1. Upgrade localstack to v3
2. Update aws-sdk-core and aws-sdk-sqs

However, I was interested in this incompatibility, so I filed this issue.
Is it possible that a similar problem occurs with AWS services other than SQS?

my guess would be xml_name in Aws::Xml::Parser::Frame.

Thank you, I agree.
When I reverted the implementation of that method, the problem was resolved. (But, I did not adopt it as a solution because I could not estimate the impact on other areas.)

Have you created an issue or feature request?

Does this mean that there is a point of contact other than this repository to report this issue?

[mullermp]

Is it possible that a similar problem occurs with AWS services other than SQS?

Possibly, but we haven't seen any reports. The query protocol is not largely used anymore.

Does this mean that there is a point of contact other than this repository to report this issue?

I meant for localstack. Is SQS JSON supported in localstack v3? If so, then your path is to upgrade. If not, create an issue or feature request with their repo and I can put a comment on it.

When I reverted the implementation of that method, the problem was resolved. (But, I did not adopt it as a solution because I could not estimate the impact on other areas.)

I would say you can use that monkey patch if it's working for you. You may as well also patch the other xml parsing changes as well just so it is complete.

[yokonao]

The query protocol is not largely used anymore.

👍
I'm going to check if the AWS services we use use the query protocol.

I meant for localstack.

I got it thanks. locastack v3 supports SQS JSON, we are working on upgrading localstack in progress.
We're having a bit of a problem because of another issue, but that's our problem.

I sincerely appreciate your response. I understand that this issue will not be fixed.

[mullermp]

Thank you. I'm sorry again that this affected you. I think it would be best to monkey patch the parser for now until you can upgrade your local stack and then finally your SDK gems.

[github-actions]

This issue is now closed. Comments on closed issues are hard for our team to see.
If you need more assistance, please open a new issue that references this one.