Github user srowen commented on the pull request:
https://github.com/apache/spark/pull/5354#issuecomment-92521916
In this case, `tachyon` includes `thrift` (shaded) and `httpclient`
(unshaded). Adding a direct dependency on `httpclient` shouldn't do anything.
Shading hasn't changed that it (unshaded) is a transitive dependency. Well,
maybe there's more to it.
But my point remains something else. If `httpclient` is needed, you can't
exclude it, right? If Spark also depends on `httpclient`, that's fine: Maven
resolves the conflict. If Maven's choice is wrong, then _Spark_ manages the
dependency in `dependencyManagement`, but, this isn't done with exclusions. For
example, this is the situation with `commons-beanutils`.
I suppose my secondary point is still that `tachyon-client` seems to pull
in lots of stuff that don't seem like things a client-side library should need,
like a web server or `hadoop-core`. I know that this is at best a Tachyon
question, but, maybe still valid to consider here since Spark is a primary
consumer of Tachyon.
(Third question, which is off topic here I suppose is, why does Tachyon
belong as a dependency of the core anyway? I suppose you could make the same
argument for Hadoop of course, but I didn't see Tachyon as so nearly essential.
I would have expected an optional module or something like YARN, since it's not
something that will be enabled by most users.)
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
