The above release files should be verified using the PGP signatures and the project release KEYS. See verification instructions for a description of using the PGP and KEYS files for verification. A SHA512 checksum is also provided as an alternative verification method.
<dependency>
<groupId>org.apache.daffodil</groupId>
<artifactId>daffodil-core_3</artifactId>
<version>4.3.0</version>
</dependency>libraryDependencies += "org.apache.daffodil" %% "daffodil-core" % "4.3.0"/etc/yum.repos.d/apache-daffodil.repo with the following content:
[apache-daffodil]
name=Apache Daffodil
baseurl=https://apache.jfrog.io/artifactory/daffodil-rpm/
enabled=1
gpgkey=https://downloads.apache.org/daffodil/KEYS
gpgcheck=1
repo_gpgcheck=0sudo dnf install apache-daffodilThis release focused on bugs fixed and performance improvements. A number of changes resulted in deprecation or backwards incompatible changes. See Deprecation/Compatibility section for more details.
This release requires Java 17 or newer, and Scala 3.3.x. It was tested with Java 17, 21, and 25, and Scala 3.3.7. We recommend Java 25 for the best performance.
[DAFFODIL-2132] - Fixed issue where delimiters could match empty bits
A schema with dfdl:initiatedContent='yes' whose dfdl:initiator lists %ES; or %WSP*; as one of its alternatives, such as "P %ES;", is now rejected with a Schema Definition Error, as the DFDL spec requires. Such schemas used to compile and parse. To migrate, set dfdl:initiatedContent='no', or drop the zero-length alternative from the initiator.
A sequence or choice group whose only initiator or terminator matches zero-length data, such as %ES; or %WSP*;, no longer counts as occupying space in the data. When such a group is last in a sequence, its separator may now be omitted on parse, and is no longer written on unparse. Data that failed with "Failed to find infix separator" now parses, and unparsed output no longer carries that trailing separator, so files compared against previously generated output will differ.
[DAFFODIL-3065] - Improve unparse performance by building the infoset tree ahead of unparsing
Unparse now builds the infoset ahead of the unparse by default. To restore the previous behavior, set the tunable infosetBuilderMode=eventDriven. Build may hold up to unparseBuildAheadWindowNodes infoset nodes in memory ahead of the unparse.
[DAFFODIL-3094] - Enforce separatorSuppressionPolicy for absent occurrences in separated sequences
Discriminators are not used when unparsing and the potentially trailing definition does not exclude them, so an optional element with a discriminator is now potentially trailing. Under trailingEmpty and trailingEmptyStrict its trailing zero-length occurrence is now suppressed with its separator when unparsing, where it was previously written. Schemas that relied on the old output must be updated.
The following dependencies have been added, updated, or removed: