[
https://issues.apache.org/jira/browse/IGNITE-20768?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kirill Sizov updated IGNITE-20768:
-----------------------------------
Description:
Investigation during the implementation of IGNITE-20609 shows that no one fails
when we have a mistake in the patch.
The first place not covered by tests in the code:
{code:title=PartitionReplicaListener.java}
if (lastCommitTimestamp != null || row != null) {
TimedBinaryRowMessageBuilder rowMsgBldr =
MSG_FACTORY.timedBinaryRowMessage();
if (lastCommitTimestamp != null) {
rowMsgBldr.timestamp(lastCommitTimestamp.longValue());
}
if (row != null) {
rowMsgBldr.binaryRowMessage(binaryRowMessage(row));
}
bldr.messageRowToUpdate(rowMsgBldr.build());
}
{code}
We need to add a test for batch operations that has both fields (not nulles):
_lastCommitTimestamp_ and _row_.
The second
{code:title=StorageUpdateHandler.java}
performStorageCleanupIfNeeded(txId, rowId, entry.getValue() == null ? null :
entry.getValue().commitTimestamp());
{code}
We need to add a test for batch operations where `entry.getValue()` is null.
was:
Investigation during the implementation of IGNITE-20609 shows that no one fails
when we have a mistace in the patch.
The first place not covered by tests in the code:
{code:title=PartitionReplicaListener.java}
if (lastCommitTimestamp != null || row != null) {
TimedBinaryRowMessageBuilder rowMsgBldr =
MSG_FACTORY.timedBinaryRowMessage();
if (lastCommitTimestamp != null) {
rowMsgBldr.timestamp(lastCommitTimestamp.longValue());
}
if (row != null) {
rowMsgBldr.binaryRowMessage(binaryRowMessage(row));
}
bldr.messageRowToUpdate(rowMsgBldr.build());
}
{code}
We need to add a test for batch operations that has both fields (not nulles):
_lastCommitTimestamp_ and _row_.
The second
{code:title=StorageUpdateHandler.java}
performStorageCleanupIfNeeded(txId, rowId, entry.getValue() == null ? null :
entry.getValue().commitTimestamp());
{code}
We need to add a test for batch operations where `entry.getValue()` is null.
> Test coverege of batch operations
> ---------------------------------
>
> Key: IGNITE-20768
> URL: https://issues.apache.org/jira/browse/IGNITE-20768
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Assignee: Kirill Sizov
> Priority: Major
> Labels: ignite-3
>
> Investigation during the implementation of IGNITE-20609 shows that no one
> fails when we have a mistake in the patch.
> The first place not covered by tests in the code:
> {code:title=PartitionReplicaListener.java}
> if (lastCommitTimestamp != null || row != null) {
> TimedBinaryRowMessageBuilder rowMsgBldr =
> MSG_FACTORY.timedBinaryRowMessage();
> if (lastCommitTimestamp != null) {
> rowMsgBldr.timestamp(lastCommitTimestamp.longValue());
> }
> if (row != null) {
> rowMsgBldr.binaryRowMessage(binaryRowMessage(row));
> }
> bldr.messageRowToUpdate(rowMsgBldr.build());
> }
> {code}
> We need to add a test for batch operations that has both fields (not nulles):
> _lastCommitTimestamp_ and _row_.
> The second
> {code:title=StorageUpdateHandler.java}
> performStorageCleanupIfNeeded(txId, rowId, entry.getValue() == null ? null :
> entry.getValue().commitTimestamp());
> {code}
> We need to add a test for batch operations where `entry.getValue()` is null.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
