[
https://issues.apache.org/jira/browse/TINKERPOP-1402?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15413578#comment-15413578
]
Bryn Cooke commented on TINKERPOP-1402:
---------------------------------------
I like the onMapper idea and also agree the registry method is redundant.
Trying to unify the serialization apis would seem to be a pretty tall order,
onMapper gives implementations the required flexibility. I think that the
registry should be deprecated.
> Impossible for graph implementations to provide a class resolver for Gryo IO
> ----------------------------------------------------------------------------
>
> Key: TINKERPOP-1402
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1402
> Project: TinkerPop
> Issue Type: Improvement
> Components: structure
> Affects Versions: 3.2.1
> Reporter: Bryn Cooke
> Priority: Critical
>
> As far as I can tell there is no way for a graph implementation to specify a
> classresolver for the following code:
> {noformat}g.io(IoCore.gryo()).writer().create(){noformat}
> The problem is that inside the graph implementation we need to be able to do
> this:
> {noformat}
> public <I extends Io> I io(final Io.Builder<I> builder) {
> Io io =
> builder.graph(this).registry(MyRegistry.INSTANCE).classResolver(MyClassReolver.INSTANCE).create();
> {noformat}
> but only supplying a registry is supported.
> Other solutions could be to design GryoIo for extension so that it can be
> wrapped or to change the signature of Graph#io to:
> {noformat}
> public default Io io(final Io.Builder<I> builder)
> {noformat}
> I would probably go for the signature change, so the graph is responsible for
> deciding the implementation that is returned.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
