Github user vasia commented on the pull request:
https://github.com/apache/flink/pull/847#issuecomment-121025189
Hi @andralungu,
do you mean support for POJOs as vertex / edge values?
I guess that's a limitation we can't easily overcome, I agree.
Still though, a nicely designed `fromCsv()` method would simplify the
common case.
As for the examples, I don't like what they currently look like in this PR
either. However, that's not a problem of `fromCsv()`. The if-block can be
easily simplified by changing `getDefaultEdgeDataSet` to `getDefaultGraph`. The
else-block looks longer because of the mapper, which, in the current examples
is in the main method.
What I think is quite problematic, is the `types()` methods. Ideally, we
would have the following:
1. `types(K)` : no vertex value, no edge value
2. `types(K, VV)`: no edge value
3. `types(K, EV)`: no vertex value
4. `types(K, VV, EV)`: both vertex and edge values are present
However, because of type erasure, we can't have both 2 and 3. The current
implementation (having separate `typesEdges` and `typesVertices`) means that
both should always be called, even if not necessary. Another way would be to
give 2 and 3 different names... So far I haven't been able to come up with a
nice solution. Ideas?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
