The Glacier client can create, delete, read, and list vaults.

… code for clarity.

Annotations removed with:

find -name \*.java | xargs sed -i /@author/d

Empty Javadoc removed with multiple iterations of:

find -name \*.java | xargs sed -i -z 's/ \*\n \*\// \*\//'
find -name \*.java | xargs sed -i -z 's/ \* \n \*\// \*\//'
find -name \*.java | xargs sed -i -z 's/\/\*\*\n \*\/\n//'

…eturn false on any IllegalArgumentException.

Now the live tests are not conditioned by the current state of the
account used.

The TreeHash class stores two HashCodes, the linear and the
tree HashCode. The TreeHash class also has an static inner class
named Hasher, which allows us to calculate the hashes.

Both Hashes (tree and linear) are needed for the upload archive
and the multipart upload operations. When completing a multipart
upload operation we also need to calculate the tree hash of the
individual part's hashes.

Also ensure that we close the InputStream.

Now the Glacier client supports upload and delete archive
operations.

An static TestUtils class has been made for the archive operations
tests. This class allows us to build payloads and build ByteSources.

On the last commit we added TestUtils class. On this commit
the VaultNameValidator and the ThreeHashTest tests make use
of it.

GlacierClientMockTest was not using GlacierHeaders references.
This commit fixes this problem too.

In addition, the AWSRequestSignerV4, the TreeHash and the
VaultNameValidator classes are now final.

GlacierClientMockTest was growing uncontrollably. This commit
is an attempt to fix this problem.

The first measure taken moves the duplicated strings to constants
and the JSON response bodies to resource files. In addition, some
repeated code was moved to static classes, where they can be reused.

We also use @BeforeTest and @AfterTest to make instances of the
MockWebServer and shut them down for each test.

Now the Glacier client supports initiateMultipartUpload and
uploadPart operations.

Now the validators use checkArgument instead of throwing an
exception by calling an static method.

There were some missing validators on the Async client, these
were fixed too.

Lastly, TreeHash inner class was removed. Now the factory methods
are in the TreeHash class.

Now the Glacier client supports completeMultipartUpload and i
abortMultipartUpload operations.

Use these annotations to initialize before every test.  BeforeTest and
AfterTest initialize once before a single test in this suite.

Added listParts and listMultipartUploads operations to the Glacier
client.

For each test, we were forcing the server to discard the content of
the request body on @BeforeMethod. We need to do this on tests with a
huge body, but it may interfere with other tests which actually use
the body data.  To fix this problem, we moved the body limit
restriction to the tests that require it.

Now the Glacier client supports the initiateJob operation.

The binder classes were making unnecessary checks. They have
been cleaned up.

This commit addresses some problems found in the code:

- Now VaultMetadata returns a copy of the creation date instead
of their own instance.

- Added a checkNotNull for PaginatedVaultCollection iterable.

- Added a test for listVaults with an empty list of vaults.

Now Glacier client supports the describeJob operation.

Now the Glacier client supports the listJobs operation.

Now the Glacier client can retrieve data.

This commit demonstrates a few instances where assertj yields more
informative error messages than testng assertEquals and assertTrue.
Note that we could replace all testng asserts with assertj.

Every project was using undefined properties in their pom.xml.
A default value for test.aws.identity and test.aws.credential was
added.

testng assertions have been replaced with assertj
assertions.

The uploadPart method now returns HashCode instead
of String.

Test groups have been added. Also, the live tests were
refactored to reduce the number of RPCs.

Now the job status is stored as an enum instead of
a String. This is useful when comparing the status to
check if a job is completed, failed or still in progress.

The Amazon Glacier BlobStore skeleton has been added.

The completeMultipartUpload operation is taking the archive size
parameter in MB but it should be bytes.

The code related to the MultiPartUpload strategy has been added.
MultiPart uploads use an upload strategy (e.g. sequential vs parallel)
and also a slicing strategy to split the payload in different parts.

The long live tests have been added. These tests takes several
hours and will leave traces in your AWS account.

List containers and remove blob operations have
been added.

The polling strategy interface for job operations
and a simple implementation class have been added.
This implementation waits an initial time and then
polls at regular intervals.

Now the BlobStore abstraction supports the getBlob Operation

Now the BlobStore abstraction supports the list Operation

Now the BlobStore abstraction supports the blobExists and the
blobMetadata operations.

Added a proper implementation to the Clear vault strategy

The ClearVaultStrategy now ignores ResourceNotFound exceptions.

deleteContainer now tries to delete the vault first to avoid long waits.
If the delete request fails, retries it every 24 hours.

Now the BlobStore and the strategy implementations are annotated with @singleton.

Added a small documentation to the blobstore classes.

Follow-on from 95d9b00

* updating Javadocs
* minor style changes

Follow-up to 8fba1f2

The missing getBlob implementation has been added.

Added more javadoc documentation. This covers the basics of the
provider.

…Quietly.

Not implemented for Glacier.

Not implemented for Glacier.

…-rc1

Not yet implemented

It is currently throwing UnsupportedOperationException exception.

Found via error-prone.

Not yet implemented.

Enable for Glacier.

jclouds should not assume a particular endpoint when running glacier
tests, as it's possible that a glacier emulator is used (e.g.
glacier-proxy).

Glacier has an initial wait parameter, which is used to check on the
progress of a job. By default, it is set to 3 hours. This may work
well against existing AWS Glacier interface, however, against a
Glacier emulator the timeout should be overriden. The change allows
setting the timeout to an arbitrary shorter value with
-Dtest.glacier.initial-wait.

This can cause thread visibility issues.  Found via error-prone.

- from 2.2.0-SNAPSHOST to 2.2.0-SNAPSHOT

This project, the aws-lab version of Apache JClouds, share the exact build instructions as the primary Apache JClouds project with all its modules.
Apache JClouds is shifting its strategy in handling OSGi configuration. Instead of using the Maven Bundle Plugin, a wrapper of the BND plugin, the BND plugin gets used directly.
- Remove the OSGi configuration from each module. The configuration gets served to the BND through dedicated configuration / bnd files.
- Onboard bnd-configuration files, one per module.

Ignore bnd files in rat plugin

The JClouds project module upgrades the okhttp server library and related dependencies such as mockwebserver from 2.2.0 to 3.14.9. The vendor switched the groupId declaration from com.squareup.okhttp to com.squareup.okhttp3.

Adjust imports and api calls for newer okhttp vers

This addresses a critical CVE:

https://logging.apache.org/log4j/2.x/security.html

…moved' into promoted-glacier