Hey JB, I don't think we're ever using e.g. *Lists.newArrayList()* without the diamond syntax in the codebase, so it's typically always *List<String> list = Lists.newArrayList()*. So I wonder how much of an issue that actually is? Do you have examples in the codebase that don't use the diamond syntax and is it worth updating the entire codebase?
to loosen the Lists/Maps/... enforcement in spotless) FYI this is being enforced by checkstyle and was first introduced by https://github.com/apache/iceberg/pull/3689 in order to have a consistent style around collection instantiation. Thanks Eduard On Thu, Oct 24, 2024 at 8:20 AM Jean-Baptiste Onofré <j...@nanthrax.net> wrote: > Hi folks, > > We are using Guava for different "utils" methods. Especially, we are > using Guava to create lists and maps. For instance, we do (and we > force the use of): > > List myList = Lists.newArrayList(); > > or > > Map myMap = Maps.newHashMap(); > > If it was a good idea up to JDK7, these methods are now unnecessary, > and the "regular" approach should be preferred in order to cleanly use > the diamond syntax > ( > https://docs.oracle.com/javase/tutorial/java/generics/genTypeInference.html#type-inference-instantiation > ). > > The Guava Javadoc actually clearly states this: > > https://github.com/google/guava/blob/v33.3.1/guava/src/com/google/common/collect/Lists.java#L79 > > So, I propose: > - to loosen the Lists/Maps/... enforcement in spotless) > - to refactore Iceberg to use JDK list and map constructors instead of > Guava Lists/Maps/... > > I started the change, but as it's a pretty large one, I would like > your feedback before completing the PR. > > Regards > JB >
