Merge pull request #68802 from Avogar/consistent-dynamic-structure

Make dynamic structure selection more consistent
This commit is contained in:
Kruglov Pavel 2024-08-24 11:53:17 +00:00 committed by GitHub
commit 293821a186
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -1181,13 +1181,14 @@ void ColumnDynamic::takeDynamicStructureFromSourceColumns(const Columns & source
/// Check if the number of all dynamic types exceeds the limit.
if (!canAddNewVariants(0, all_variants.size()))
{
/// Create list of variants with their sizes and sort it.
std::vector<std::pair<size_t, DataTypePtr>> variants_with_sizes;
/// Create a list of variants with their sizes and names and then sort it.
std::vector<std::tuple<size_t, String, DataTypePtr>> variants_with_sizes;
variants_with_sizes.reserve(all_variants.size());
for (const auto & variant : all_variants)
{
if (variant->getName() != getSharedVariantTypeName())
variants_with_sizes.emplace_back(total_sizes[variant->getName()], variant);
auto variant_name = variant->getName();
if (variant_name != getSharedVariantTypeName())
variants_with_sizes.emplace_back(total_sizes[variant_name], variant_name, variant);
}
std::sort(variants_with_sizes.begin(), variants_with_sizes.end(), std::greater());
@ -1196,14 +1197,14 @@ void ColumnDynamic::takeDynamicStructureFromSourceColumns(const Columns & source
result_variants.reserve(max_dynamic_types + 1); /// +1 for shared variant.
/// Add shared variant.
result_variants.push_back(getSharedVariantDataType());
for (const auto & [size, variant] : variants_with_sizes)
for (const auto & [size, variant_name, variant_type] : variants_with_sizes)
{
/// Add variant to the resulting variants list until we reach max_dynamic_types.
if (canAddNewVariant(result_variants.size()))
result_variants.push_back(variant);
result_variants.push_back(variant_type);
/// Add all remaining variants into shared_variants_statistics until we reach its max size.
else if (new_statistics.shared_variants_statistics.size() < Statistics::MAX_SHARED_VARIANT_STATISTICS_SIZE)
new_statistics.shared_variants_statistics[variant->getName()] = size;
new_statistics.shared_variants_statistics[variant_name] = size;
else
break;
}