You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@iceberg.apache.org by "huan233usc (via GitHub)" <gi...@apache.org> on 2026/07/02 03:52:10 UTC

[PR] feat(avro): apply column default values when reading missing fields (3/4) [iceberg-cpp]

huan233usc opened a new pull request, #800:
URL: https://github.com/apache/iceberg-cpp/pull/800

   ## What
   
   Part 3 of 4 of Iceberg v3 column default-value support (POC #731), built on the
   schema layer (#746) and the Parquet read path (#792).
   
   When a column is present in the read (table) schema but absent from an Avro data
   file — because the column was added after those rows were written — fill it with
   the column's v3 `initial-default` instead of `null`.
   
   ## Changes
   
   - **`arrow/literal_util`**: add `AppendDefaultToBuilder`, which appends a
     `Literal` once to an Arrow `ArrayBuilder` (for the row-by-row Avro decode
     paths). Like `MakeDefaultArray`, it targets the storage type for an extension
     builder (`arrow.uuid`), since `Scalar::CastTo` has no kernel for extension
     types.
   - **Avro projection** (`avro_schema_util.cc`): when a field is missing from the
     file and carries an `initial-default`, project it as
     `FieldProjection::Kind::kDefault`, mirroring the generic / Parquet paths.
   - **Avro decode** (`avro_data_util.cc`, `avro_direct_decoder.cc`): materialize
     the `kDefault` branch via `AppendDefaultToBuilder`.
   
   ## Tests
   
   - `literal_util_test`: `AppendDefaultToBuilder` appends a value and casts to the
     builder type.
   - `avro_data_test`: `AppendDatumToBuilder` fills missing required and optional
     default fields.
   - `avro_test`: end-to-end — write an Avro file with an old schema, then read it
     through `ReaderFactoryRegistry` with an evolved schema carrying defaults
     (`ReadMissingFieldsWithDefaults`).
   
   ## Stack
   
   1. #746 — schema: represent / serialize / validate (merged)
   2. #792 — read path: Parquet (merged)
   3. **this PR** — read path: Avro
   4. schema evolution: `addColumn` / `updateColumnDefault` (#793)
   
   Draft while the earlier PRs in the stack settle.
   


-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "wgtmac (via GitHub)" <gi...@apache.org>.
wgtmac commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3695115977


##########
src/iceberg/avro/avro_data_util.cc:
##########
@@ -497,6 +507,135 @@ Status AppendFieldToBuilder(const ::avro::NodePtr& avro_node,
 
 }  // namespace
 
+namespace {
+
+Result<std::shared_ptr<::arrow::Scalar>> MakeDefaultScalar(
+    const Literal& literal, const std::shared_ptr<::arrow::DataType>& builder_type) {
+  // The builder's own memory pool is not exposed, so the small scalar buffer uses the
+  // default pool.
+  ICEBERG_ASSIGN_OR_RAISE(std::shared_ptr<::arrow::Scalar> scalar,
+                          arrow::ToArrowScalar(literal, ::arrow::default_memory_pool()));
+
+  // For an extension builder (e.g. `arrow.uuid`) target its storage type: ToArrowScalar
+  // yields the storage scalar (fixed_size_binary(16) for uuid) and Scalar::CastTo has no
+  // kernel that targets an extension type. This mirrors MakeDefaultArray's extension
+  // handling.
+  std::shared_ptr<::arrow::DataType> target_type = builder_type;
+  if (target_type->id() == ::arrow::Type::EXTENSION) {
+    target_type = internal::checked_cast<const ::arrow::ExtensionType&>(*target_type)
+                      .storage_type();
+  }
+
+  if (!scalar->type->Equals(*target_type)) {
+    ICEBERG_ARROW_ASSIGN_OR_RETURN(scalar, scalar->CastTo(target_type));
+  }
+  return scalar;
+}
+
+Status PrepareStructDefaultScalars(std::span<FieldProjection> projections,
+                                   ::arrow::ArrayBuilder* builder);
+
+// Recurse into whatever nested builder this projection describes, so a default cached for
+// a struct field is found at any nesting depth (e.g. `list<list<struct<...>>>`) instead
+// of only when the collection's child is an immediate struct.
+Status PrepareNestedDefaultScalars(FieldProjection& projection,
+                                   ::arrow::ArrayBuilder* builder) {
+  if (projection.kind != FieldProjection::Kind::kProjected ||
+      projection.children.empty()) {
+    return {};
+  }
+
+  switch (builder->type()->id()) {
+    case ::arrow::Type::STRUCT:
+      return PrepareStructDefaultScalars(projection.children, builder);
+    case ::arrow::Type::LIST: {
+      // List projections store a single child for the element.
+      auto* list_builder = internal::checked_cast<::arrow::ListBuilder*>(builder);
+      return PrepareNestedDefaultScalars(projection.children[0],
+                                         list_builder->value_builder());
+    }
+    case ::arrow::Type::LARGE_LIST: {
+      auto* list_builder = internal::checked_cast<::arrow::LargeListBuilder*>(builder);
+      return PrepareNestedDefaultScalars(projection.children[0],
+                                         list_builder->value_builder());
+    }
+    case ::arrow::Type::MAP: {
+      auto* map_builder = internal::checked_cast<::arrow::MapBuilder*>(builder);
+      if (projection.children.size() >= 1) {
+        ICEBERG_RETURN_UNEXPECTED(PrepareNestedDefaultScalars(
+            projection.children[0], map_builder->key_builder()));
+      }
+      if (projection.children.size() >= 2) {
+        ICEBERG_RETURN_UNEXPECTED(PrepareNestedDefaultScalars(
+            projection.children[1], map_builder->item_builder()));
+      }
+      return {};
+    }
+    default:
+      return {};
+  }
+}
+
+Status PrepareStructDefaultScalars(std::span<FieldProjection> projections,
+                                   ::arrow::ArrayBuilder* builder) {
+  auto* struct_builder = internal::checked_cast<::arrow::StructBuilder*>(builder);
+  if (static_cast<size_t>(struct_builder->num_fields()) != projections.size()) {
+    return InvalidArgument(
+        "Inconsistent number of struct builder fields ({}) and projections ({})",
+        struct_builder->num_fields(), projections.size());
+  }
+
+  for (size_t i = 0; i < projections.size(); ++i) {
+    auto& field_projection = projections[i];
+    auto* field_builder = struct_builder->field_builder(static_cast<int>(i));
+
+    if (field_projection.kind == FieldProjection::Kind::kDefault) {
+      if (field_projection.attributes != nullptr) {

Review Comment:
   `attributes != nullptr` does not guarantee that `default_scalar` is prepared now that this container may hold other Avro attributes. If another attribute creates the container first, this skips preparation and the append path falls back to rebuilding the scalar per row. Please reuse/cast the existing `AvroExtraAttributes`, check `default_scalar` instead, and add a test with a pre-existing empty container.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "wgtmac (via GitHub)" <gi...@apache.org>.
wgtmac commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3649740719


##########
src/iceberg/avro/avro_data_util.cc:
##########
@@ -497,6 +507,128 @@ Status AppendFieldToBuilder(const ::avro::NodePtr& avro_node,
 
 }  // namespace
 
+namespace {
+
+Result<std::shared_ptr<::arrow::Scalar>> MakeDefaultScalar(
+    const Literal& literal, const std::shared_ptr<::arrow::DataType>& builder_type) {
+  // The builder's own memory pool is not exposed, so the small scalar buffer uses the
+  // default pool.
+  ICEBERG_ASSIGN_OR_RAISE(std::shared_ptr<::arrow::Scalar> scalar,
+                          arrow::ToArrowScalar(literal, ::arrow::default_memory_pool()));
+
+  // For an extension builder (e.g. `arrow.uuid`) target its storage type: ToArrowScalar
+  // yields the storage scalar (fixed_size_binary(16) for uuid) and Scalar::CastTo has no
+  // kernel that targets an extension type. This mirrors MakeDefaultArray's extension
+  // handling.
+  std::shared_ptr<::arrow::DataType> target_type = builder_type;
+  if (target_type->id() == ::arrow::Type::EXTENSION) {
+    target_type = internal::checked_cast<const ::arrow::ExtensionType&>(*target_type)
+                      .storage_type();
+  }
+
+  if (!scalar->type->Equals(*target_type)) {
+    ICEBERG_ARROW_ASSIGN_OR_RETURN(scalar, scalar->CastTo(target_type));
+  }
+  return scalar;
+}
+
+Status PrepareDefaultScalars(std::span<FieldProjection> projections,
+                             ::arrow::ArrayBuilder* builder) {
+  auto* struct_builder = internal::checked_cast<::arrow::StructBuilder*>(builder);
+  if (static_cast<size_t>(struct_builder->num_fields()) != projections.size()) {
+    return InvalidArgument(
+        "Inconsistent number of struct builder fields ({}) and projections ({})",
+        struct_builder->num_fields(), projections.size());
+  }
+
+  for (size_t i = 0; i < projections.size(); ++i) {
+    auto& field_projection = projections[i];
+    auto* field_builder = struct_builder->field_builder(static_cast<int>(i));
+
+    if (field_projection.kind == FieldProjection::Kind::kDefault) {
+      if (dynamic_cast<const AvroDefaultAttributes*>(field_projection.attributes.get()) !=
+          nullptr) {
+        continue;
+      }
+      auto attrs = std::make_shared<AvroDefaultAttributes>();
+      ICEBERG_ASSIGN_OR_RAISE(attrs->scalar,
+                              MakeDefaultScalar(std::get<Literal>(field_projection.from),
+                                                field_builder->type()));
+      field_projection.attributes = std::move(attrs);
+      continue;
+    }
+
+    if (field_projection.kind != FieldProjection::Kind::kProjected ||
+        field_projection.children.empty()) {
+      continue;
+    }
+
+    switch (field_builder->type()->id()) {
+      case ::arrow::Type::STRUCT:
+        ICEBERG_RETURN_UNEXPECTED(
+            PrepareDefaultScalars(field_projection.children, field_builder));
+        break;
+      case ::arrow::Type::LIST: {
+        // List projections store a single child for the element. Defaults only appear
+        // on nested struct fields of that element.
+        auto* list_builder = internal::checked_cast<::arrow::ListBuilder*>(field_builder);
+        auto& element_projection = field_projection.children[0];
+        if (element_projection.kind == FieldProjection::Kind::kProjected &&

Review Comment:
   Please recurse through nested list and map builders, not just an immediate struct. For example, a default under `list<list<struct<...>>>` is not cached and is rebuilt for every element. Please add a nested-collection test as well.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "wgtmac (via GitHub)" <gi...@apache.org>.
wgtmac merged PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800


-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "wgtmac (via GitHub)" <gi...@apache.org>.
wgtmac commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3672850996


##########
src/iceberg/avro/avro_data_util_internal.h:
##########
@@ -19,14 +19,48 @@
 
 #pragma once
 
+#include <memory>
+
 #include <arrow/array/builder_base.h>
+#include <arrow/scalar.h>
 #include <avro/GenericDatum.hh>
 
 #include "iceberg/arrow/metadata_column_util_internal.h"
+#include "iceberg/expression/literal.h"
 #include "iceberg/schema_util.h"
 
 namespace iceberg::avro {
 
+/// \brief Cached Arrow scalar for a `kDefault` field projection.
+///
+/// Materialized once (see `PrepareDefaultScalars`) so row-by-row Avro decode only
+/// needs `AppendScalar` instead of repeating `ToArrowScalar` / `CastTo` per row.
+struct AvroDefaultAttributes : FieldProjection::ExtraAttributes {

Review Comment:
   Could this follow the `ParquetExtraAttributes` pattern and become `AvroExtraAttributes` with a `default_scalar` member? `FieldProjection` has a single attributes slot, so separate subclasses for future Avro attributes would not compose. A single Avro-specific container would also let the append path use `checked_cast` instead of a per-row `dynamic_cast`.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "huan233usc (via GitHub)" <gi...@apache.org>.
huan233usc commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3678269503


##########
src/iceberg/avro/avro_data_util_internal.h:
##########
@@ -19,14 +19,48 @@
 
 #pragma once
 
+#include <memory>
+
 #include <arrow/array/builder_base.h>
+#include <arrow/scalar.h>
 #include <avro/GenericDatum.hh>
 
 #include "iceberg/arrow/metadata_column_util_internal.h"
+#include "iceberg/expression/literal.h"
 #include "iceberg/schema_util.h"
 
 namespace iceberg::avro {
 
+/// \brief Cached Arrow scalar for a `kDefault` field projection.
+///
+/// Materialized once (see `PrepareDefaultScalars`) so row-by-row Avro decode only
+/// needs `AppendScalar` instead of repeating `ToArrowScalar` / `CastTo` per row.
+struct AvroDefaultAttributes : FieldProjection::ExtraAttributes {

Review Comment:
   Done — renamed to `AvroExtraAttributes` with a `default_scalar` member, mirroring `ParquetExtraAttributes`. Since Avro now has a single attributes container, the append path uses `checked_cast` instead of a per-row `dynamic_cast`, and future Avro attributes can be added to the same struct rather than as competing subclasses.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "huan233usc (via GitHub)" <gi...@apache.org>.
huan233usc commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3653197117


##########
src/iceberg/avro/avro_data_util.cc:
##########
@@ -497,6 +507,128 @@ Status AppendFieldToBuilder(const ::avro::NodePtr& avro_node,
 
 }  // namespace
 
+namespace {
+
+Result<std::shared_ptr<::arrow::Scalar>> MakeDefaultScalar(
+    const Literal& literal, const std::shared_ptr<::arrow::DataType>& builder_type) {
+  // The builder's own memory pool is not exposed, so the small scalar buffer uses the
+  // default pool.
+  ICEBERG_ASSIGN_OR_RAISE(std::shared_ptr<::arrow::Scalar> scalar,
+                          arrow::ToArrowScalar(literal, ::arrow::default_memory_pool()));
+
+  // For an extension builder (e.g. `arrow.uuid`) target its storage type: ToArrowScalar
+  // yields the storage scalar (fixed_size_binary(16) for uuid) and Scalar::CastTo has no
+  // kernel that targets an extension type. This mirrors MakeDefaultArray's extension
+  // handling.
+  std::shared_ptr<::arrow::DataType> target_type = builder_type;
+  if (target_type->id() == ::arrow::Type::EXTENSION) {
+    target_type = internal::checked_cast<const ::arrow::ExtensionType&>(*target_type)
+                      .storage_type();
+  }
+
+  if (!scalar->type->Equals(*target_type)) {
+    ICEBERG_ARROW_ASSIGN_OR_RETURN(scalar, scalar->CastTo(target_type));
+  }
+  return scalar;
+}
+
+Status PrepareDefaultScalars(std::span<FieldProjection> projections,
+                             ::arrow::ArrayBuilder* builder) {
+  auto* struct_builder = internal::checked_cast<::arrow::StructBuilder*>(builder);
+  if (static_cast<size_t>(struct_builder->num_fields()) != projections.size()) {
+    return InvalidArgument(
+        "Inconsistent number of struct builder fields ({}) and projections ({})",
+        struct_builder->num_fields(), projections.size());
+  }
+
+  for (size_t i = 0; i < projections.size(); ++i) {
+    auto& field_projection = projections[i];
+    auto* field_builder = struct_builder->field_builder(static_cast<int>(i));
+
+    if (field_projection.kind == FieldProjection::Kind::kDefault) {
+      if (dynamic_cast<const AvroDefaultAttributes*>(field_projection.attributes.get()) !=
+          nullptr) {
+        continue;
+      }
+      auto attrs = std::make_shared<AvroDefaultAttributes>();
+      ICEBERG_ASSIGN_OR_RAISE(attrs->scalar,
+                              MakeDefaultScalar(std::get<Literal>(field_projection.from),
+                                                field_builder->type()));
+      field_projection.attributes = std::move(attrs);
+      continue;
+    }
+
+    if (field_projection.kind != FieldProjection::Kind::kProjected ||
+        field_projection.children.empty()) {
+      continue;
+    }
+
+    switch (field_builder->type()->id()) {
+      case ::arrow::Type::STRUCT:
+        ICEBERG_RETURN_UNEXPECTED(
+            PrepareDefaultScalars(field_projection.children, field_builder));
+        break;
+      case ::arrow::Type::LIST: {
+        // List projections store a single child for the element. Defaults only appear
+        // on nested struct fields of that element.
+        auto* list_builder = internal::checked_cast<::arrow::ListBuilder*>(field_builder);
+        auto& element_projection = field_projection.children[0];
+        if (element_projection.kind == FieldProjection::Kind::kProjected &&

Review Comment:
   Done — the recursion now dispatches on the builder type (struct/list/large_list/map) instead of only descending when the collection child is an immediate struct, so a default under `list<list<struct<...>>>` gets its scalar prepared once at any depth. Added a nested-collection test asserting the scalar is prepared, plus an end-to-end `list<list<struct>>` decode test (the former fails without the change).



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "huan233usc (via GitHub)" <gi...@apache.org>.
huan233usc commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3631817298


##########
src/iceberg/avro/avro_data_util.cc:
##########
@@ -497,6 +506,29 @@ Status AppendFieldToBuilder(const ::avro::NodePtr& avro_node,
 
 }  // namespace
 
+Status AppendDefaultToBuilder(const Literal& literal, ::arrow::ArrayBuilder* builder) {
+  // The builder's own memory pool is not exposed, so the small scalar buffer uses the
+  // default pool.
+  ICEBERG_ASSIGN_OR_RAISE(std::shared_ptr<::arrow::Scalar> scalar,
+                          arrow::ToArrowScalar(literal, ::arrow::default_memory_pool()));
+

Review Comment:
   Fixed in 0e4c6a40: added `PrepareDefaultScalars` to cache a cast Arrow scalar in `FieldProjection::attributes` (`AvroDefaultAttributes`) when the Avro reader initializes; per-row default fills now only call `AppendScalar`.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "wgtmac (via GitHub)" <gi...@apache.org>.
wgtmac commented on PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#issuecomment-5162086971

   Thanks for taking feedback quickly!


-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org


Re: [PR] feat(avro): apply column default values when reading missing fields [iceberg-cpp]

Posted by "Copilot (via GitHub)" <gi...@apache.org>.
Copilot commented on code in PR #800:
URL: https://github.com/apache/iceberg-cpp/pull/800#discussion_r3627414694


##########
src/iceberg/avro/avro_data_util.cc:
##########
@@ -497,6 +506,29 @@ Status AppendFieldToBuilder(const ::avro::NodePtr& avro_node,
 
 }  // namespace
 
+Status AppendDefaultToBuilder(const Literal& literal, ::arrow::ArrayBuilder* builder) {
+  // The builder's own memory pool is not exposed, so the small scalar buffer uses the
+  // default pool.
+  ICEBERG_ASSIGN_OR_RAISE(std::shared_ptr<::arrow::Scalar> scalar,
+                          arrow::ToArrowScalar(literal, ::arrow::default_memory_pool()));
+

Review Comment:
   `AppendDefaultToBuilder` constructs an Arrow scalar (and may cast it) for every appended row. When a column is missing from the Avro file, this path runs once per row per defaulted field and can become a hotspot (especially for string/binary/fixed/uuid/decimal where `ToArrowScalar` allocates buffers and `CastTo` may allocate).
   
   Consider precomputing the (already-cast) scalar once per projected field (e.g., stash it in `FieldProjection::attributes` or in the decoder/read context) and reusing it for each row append, so the per-row work is just `AppendScalar`.



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org