[
https://issues.apache.org/jira/browse/IGNITE-23898?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Iurii Gerzhedovich updated IGNITE-23898:
----------------------------------------
Description:
Under IGNITE-23727, the method `map` has been moved from the RowHandler class
into the RowFactory class, which is logically right because the method
producing a new row should be placed in a factory instead of a handler. Also,
it helps to avoid calculating schema for each method invocation or passing it
as a parameter each time.
Right now we have another similar method in RowHandler named `concat`. Let's
also move it into the RowFactory and calculate scham just once.
Also, let's do a little refactoring for existed method
ModifyNode.getMappedRowSchema, need to extract a new static method for mapping
into RowSchema class like a :
{code:java}
class RowSchema {
static RowSchema map(RowSchema schema, int[] mapping) {
List<TypeSpec> fields = schema.fields();
Builder builder = builder();
for (int i : mapping) {
TypeSpec typeSpec = fields.get(i);
builder.addField(typeSpec);
}
return builder.build();
}
} {code}
was:
Under IGNITE-23727, the method `map` has been moved from the RowHandler class
into the RowFactory class, which is logically right because the method
producing a new row should be placed in a factory instead of a handler. Also,
it helps to avoid calculating schema for each method invocation or passing it
as a parameter each time.
Right now we have another similar method in RowHandler named `concat`. Let's
also move it into the RowFactory and calculate scham just once.
Also, let's do a little refactoring for existed method
ModifyNode.getMappedRowSchema, need to extract a new static method for mapping
into RowSchema class like a :
{code:java}
class RowSchema {
static RowSchema map(RowSchema schema, int[] mapping) {
List<TypeSpec> fields = schema.fields();
Builder builder = builder();
for (int i : mapping)
{ TypeSpec typeSpec = fields.get(i); builder.addField(typeSpec); }
return builder.build();
}
{code}
{{}}
> Sql. Calculate schema for concat two rows just once
> ---------------------------------------------------
>
> Key: IGNITE-23898
> URL: https://issues.apache.org/jira/browse/IGNITE-23898
> Project: Ignite
> Issue Type: Bug
> Components: sql
> Reporter: Iurii Gerzhedovich
> Priority: Major
>
> Under IGNITE-23727, the method `map` has been moved from the RowHandler class
> into the RowFactory class, which is logically right because the method
> producing a new row should be placed in a factory instead of a handler. Also,
> it helps to avoid calculating schema for each method invocation or passing it
> as a parameter each time.
> Right now we have another similar method in RowHandler named `concat`. Let's
> also move it into the RowFactory and calculate scham just once.
> Also, let's do a little refactoring for existed method
> ModifyNode.getMappedRowSchema, need to extract a new static method for
> mapping into RowSchema class like a :
> {code:java}
> class RowSchema {
> static RowSchema map(RowSchema schema, int[] mapping) {
> List<TypeSpec> fields = schema.fields();
> Builder builder = builder();
> for (int i : mapping) {
> TypeSpec typeSpec = fields.get(i);
> builder.addField(typeSpec);
> }
> return builder.build();
> }
> } {code}
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
