Github user okram commented on the issue:
https://github.com/apache/tinkerpop/pull/535
Either works, but I'm leaning more towards (a). I would do something like:
Assume the traversal:
`g.V().out().values('name').count()`
The test algorithm would:
1. I
Github user dkuppitz commented on the issue:
https://github.com/apache/tinkerpop/pull/535
So do you suggest a) to remove the assertions that check the metrics for
specific steps or b) to find the indexes of all steps that were present before
strategies were applied?
---
If your proj
Github user okram commented on the issue:
https://github.com/apache/tinkerpop/pull/535
This is not the way that this should be done. We shouldn't remove any
strategies. What we should do is generalize the `ProfileTests` so they are not
so specific about step indexes and the like. For
Github user dkuppitz commented on the issue:
https://github.com/apache/tinkerpop/pull/535
The setup method now removes `LazyBarrierStrategy` and all provider
specific strategies. We can't remove all strategies, since certain test suites
use a different set of extra strategies that are
Github user okram commented on the issue:
https://github.com/apache/tinkerpop/pull/535
I was thinking more about this specific problem. The real problem is
`ProfileTest`, not `LazyBarrierStrategy`. We need to generalize all the
`ProfileTest` test cases such that they not be concerned
Github user dkuppitz commented on the issue:
https://github.com/apache/tinkerpop/pull/535
That was my plan. kinda. I was going to open a new ticket to get rid of
`is.testing` altogether.
---
If your project is set up for it, you can reply to this email and have your
reply appear on G
Github user okram commented on the issue:
https://github.com/apache/tinkerpop/pull/535
There are numerous areas where `is.testing` is used for strategies. We
shouldn't just do this for one of them and push a PR. We should overhaul the
entire system so we don't have some parts of the c
