pitrou commented on code in PR #13537: URL: https://github.com/apache/arrow/pull/13537#discussion_r920246357
########## cpp/src/arrow/engine/substrait/options.h: ########## @@ -0,0 +1,55 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// This API is EXPERIMENTAL. + +#pragma once + +namespace arrow { +namespace engine { + +/// How strictly to adhere to the input structure when converting between Substrait and +/// Acero representations of a plan. This allows the user to trade conversion accuracy +/// for performance and lenience. +enum class ConversionStrictness { + /// Prevent information loss by rejecting incoming plans that use features or contain + /// metadata that cannot be exactly represented in the output format in a way that + /// will round-trip. Relations/nodes must map one-to-one. + PEDANTIC, Review Comment: "Pedantic" conveys an idea of spurious or arbitrary constraints. Is this the case here? Otherwise "strict" would sound better IMHO. ########## cpp/src/arrow/engine/substrait/options.h: ########## @@ -0,0 +1,55 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// This API is EXPERIMENTAL. + +#pragma once + +namespace arrow { +namespace engine { + +/// How strictly to adhere to the input structure when converting between Substrait and +/// Acero representations of a plan. This allows the user to trade conversion accuracy +/// for performance and lenience. +enum class ConversionStrictness { + /// Prevent information loss by rejecting incoming plans that use features or contain + /// metadata that cannot be exactly represented in the output format in a way that + /// will round-trip. Relations/nodes must map one-to-one. + PEDANTIC, + + /// When an incoming plan uses a feature that cannot be exactly represented in the + /// output format, attempt to emulate that feature as opposed to immediately + /// rejecting the plan. For example, a Substrait SortRel with a complex sort key + /// expression may be emulated using a project-order-project triple. Relations/nodes + /// will thus map one-to-many. Review Comment: ```suggestion /// may thus map one-to-many. ``` ########## cpp/src/arrow/engine/substrait/options.h: ########## @@ -0,0 +1,55 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// This API is EXPERIMENTAL. + +#pragma once + +namespace arrow { +namespace engine { + +/// How strictly to adhere to the input structure when converting between Substrait and +/// Acero representations of a plan. This allows the user to trade conversion accuracy +/// for performance and lenience. +enum class ConversionStrictness { + /// Prevent information loss by rejecting incoming plans that use features or contain + /// metadata that cannot be exactly represented in the output format in a way that + /// will round-trip. Relations/nodes must map one-to-one. + PEDANTIC, + + /// When an incoming plan uses a feature that cannot be exactly represented in the + /// output format, attempt to emulate that feature as opposed to immediately + /// rejecting the plan. For example, a Substrait SortRel with a complex sort key + /// expression may be emulated using a project-order-project triple. Relations/nodes + /// will thus map one-to-many. + PRESERVE_STRUCTURE, + + /// Attempt to prevent performance-related regressions caused by differences in how + /// operations are represented in the input and output format, by allowing for + /// optimizations that cross structural boundaries. For example, the converter may + /// collapse chains of project nodes into one. Review Comment: A more general question is: is BEST_EFFORT always more lenient than PRESERVE_STRUCTURE? In PRESERVE_STRUCTURE I see "Relations/nodes will thus map one-to-many", does BEST_EFFORT also allow the converse (relations/nodes mapping many-to-one)? ########## cpp/src/arrow/engine/substrait/options.h: ########## @@ -0,0 +1,55 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// This API is EXPERIMENTAL. + +#pragma once + +namespace arrow { +namespace engine { + +/// How strictly to adhere to the input structure when converting between Substrait and +/// Acero representations of a plan. This allows the user to trade conversion accuracy +/// for performance and lenience. +enum class ConversionStrictness { + /// Prevent information loss by rejecting incoming plans that use features or contain + /// metadata that cannot be exactly represented in the output format in a way that + /// will round-trip. Relations/nodes must map one-to-one. + PEDANTIC, + + /// When an incoming plan uses a feature that cannot be exactly represented in the + /// output format, attempt to emulate that feature as opposed to immediately + /// rejecting the plan. For example, a Substrait SortRel with a complex sort key + /// expression may be emulated using a project-order-project triple. Relations/nodes + /// will thus map one-to-many. + PRESERVE_STRUCTURE, + + /// Attempt to prevent performance-related regressions caused by differences in how + /// operations are represented in the input and output format, by allowing for + /// optimizations that cross structural boundaries. For example, the converter may + /// collapse chains of project nodes into one. Review Comment: Isn't collapsing of a chain of project nodes more of an optimization than conversion concern? The collapsed node should be strictly equivalent to the original chain AFAIU. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
