Trafficgen updates #858
Trafficgen updates #858
Conversation
|
Hmm, not sure what is going on with travis-ci but something seems off. |
|
It's a unit test failure. Try this in your local branch: It should fail. |
|
Thanks @ndokos. I looked at the output and I guess I just scrolled past that without seeing it (multiple times)... |
| trex_use_l2="n" | ||
| benchmark_run_dir="" | ||
| traffic_directions="bidirec" | ||
| traffic_directions="bidirectional" |
There was a problem hiding this comment.
Do we need to be binary compatible, and support bidirec to mean bidirectional?
There was a problem hiding this comment.
No. There are a lot more changes coming like this, we are going to be changing the entire option structure and don't have any need for backwards compatibility.
There was a problem hiding this comment.
Shall we make these changes in 0.52 instead of 0.51 so that you can accumulate a number of them into one release?
There was a problem hiding this comment.
The option structure changes are not pending real soon, that is more a forward looking viewpoint. The changes and additions in this PR are needed to deploy new functionality to users sooner rather than later.
|
Let's wait for @atheurer to review before we merge. Do we have any unit tests for |
- Use the same parameter and options as binary-search.py for better consistency.
- Lots of code rework since the trex-txrx-profile traffic generator is a bit different than previous traffic generators since most of the traffic parameters exist in the profile (on a per stream basis) instead of as CLI arguments (and applying to all streams).
… turned off
- It does not make sense to aggregate some data sets so add a flag,
get_label('skip_aggregate_label'), which will force datasets to be
ignored when aggregation is being performed.
atheurer
left a comment
atheurer
left a comment
There was a problem hiding this comment.
Everything looks good to me.
- This makes it easier to postprocess data since the requirements are lower. This includes the ability to test the postprocessor on a system that does not run the benchmark.
4 participants
No description provided.