Osumi Framework
es en eu v9.12.1 GitHub

Migration Guide: 9.10.2 → 9.11.0

Osumi Framework 9.11.0 adds native ORM support for SQL DECIMAL(p,s) columns.

The update is backward compatible and existing model definitions do not require changes.

DECIMAL Support

DECIMAL fields use OField::DECIMAL and PHP ?string properties.

#[OField(
	type: OField::DECIMAL,
	precision: 12,
	scale: 2,
	nullable: false
)]
public ?string $amount = null;

Using string prevents intermediate floating-point conversions and preserves the exact decimal value.

float and int values are not valid for DECIMAL fields.

The default precision is 10 and the default scale is 2. The maximum supported precision is 65 and the maximum scale is 30.

Both . and , are accepted as input decimal separators. The ORM always normalizes values to . and fixed scale:

"12"   → "12.00"
"12.5" → "12.50"
"12,5" → "12.50"

The ORM never performs automatic rounding. Values exceeding the configured precision or scale are rejected.

OField::DECIMAL cannot be used in #[OPK] fields.

Queries

Conditions passed to where() and count() are now validated and normalized according to the model schema.

For example, for a DECIMAL(12,2) field:

Payment::where([
	'amount' => '12,5'
]);

internally uses "12.50".

A floating-point value for the same field is rejected:

Payment::where([
	'amount' => 12.5
]);

Conditions referencing fields that do not exist in the model are also rejected.

Generators

generateModel, generateModelFrom and generateModelFromDB support DECIMAL fields.

generateModelFromDB preserves decimal precision, scale and default values as exact strings.

PHP 8.5 Compatibility

ORM schema initialization now checks ReflectionProperty::hasDefaultValue() before calling ReflectionProperty::getDefaultValue(), avoiding the PHP 8.5 deprecation notice for properties without an explicit default value.

Migration

There is no automatic migration step for this release.

Existing projects that do not use OField::DECIMAL do not need to modify their models.

Applications adopting DECIMAL should represent these values as exact strings and avoid converting them to float.