[
https://issues.apache.org/jira/browse/THRIFT-4379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16226066#comment-16226066
]
Jens Geyer commented on THRIFT-4379:
------------------------------------
In theory we could add a server call to the ThriftTest API to have the server
terminate via API. That would add some determinism. I have a patch in my
private repo that works around another race condition on slow machines, IIRC in
ruby somewhere. Not a good fix, so I really don't want to contrib that.
> Haxe "make check" requires "sleep 10" to complete
> -------------------------------------------------
>
> Key: THRIFT-4379
> URL: https://issues.apache.org/jira/browse/THRIFT-4379
> Project: Thrift
> Issue Type: Bug
> Components: Haxe - Library
> Affects Versions: 0.10.0
> Environment: docker ubuntu-xenial
> Reporter: James E. King, III
> Assignee: Oleksii Prudkyi
> Priority: Minor
>
> If you fire up the ubuntu-xenial docker container and run
> {{build/docker/script/autotools.sh}}, it will run all the unit tests
> including haxe. When the haxe tests run, they have a "sleep 10" at the end
> of each test. Something is holding on to resources that it should not be
> following the conclusion of the unit tests in order to require this sleep to
> pass. Please investigate why the sleep is there and whether it can be
> removed? This adds 30 seconds to every build unnecessarily and worse is
> indicative of a logic error in the code.
> See test/haxe/Makefile.am - that is where the sleeps originate from.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
