[
https://issues.apache.org/jira/browse/FLINK-1520?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14584586#comment-14584586
]
Andra Lungu commented on FLINK-1520:
------------------------------------
Hey [~vkalavri]],
To my knowledge, you cannot deduce the key or the value's class from the
generic <K,VV,EV>. The way I would implement fromCsv is by adding the classes
as parameters, e.g. Graph.fromCsv(edgesPath, String.class, String.class,
context). For NullValue, then, we would have a single class argument
Graph.fromCsv(edgesPath, String.class, context).
The user should know what kind of keys he/she has in there. So the extra
parameters should not be that much of a burden.
Is this what you had in mind? For the time being, I cannot see a "smarter" way
of doing it :)
The examples should be updated accordingly since they now read the edge and
vertex data sets from CSV and then use fromDataSet to produce the graph.
> Read edges and vertices from CSV files
> --------------------------------------
>
> Key: FLINK-1520
> URL: https://issues.apache.org/jira/browse/FLINK-1520
> Project: Flink
> Issue Type: New Feature
> Components: Gelly
> Reporter: Vasia Kalavri
> Assignee: Shivani Ghatge
> Priority: Minor
> Labels: easyfix, newbie
>
> Add methods to create Vertex and Edge Datasets directly from CSV file inputs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
