Download/Upload from/to S3 fails when using more than one AWS account in one script

[centic9]

I have a job which copies data from one S3 account to another, thus having to access two different S3 repositories in one ant build script.

Because caching is done based on AmazonS3Client.class, client-instances are cached even if you use different aws-key/secretkey combinations in one ant file, consequently such ant-scripts fail with access error.

Maybe we can add an option to disable this caching altogether as it is probably not a huge time-saver unless you upload a lot of small files.

Likely other types of tasks are affected as well.

[fulghum]

Hey Dominik, I agree with you that the caching isn't an important performance win. The correctness of the code is of course more important.

If you'd like to submit a pull request to disable it, we'd be happy to accept it. Otherwise, we'll add the change to our backlog.