search

HydrologicEngineeringCenter / fda-model

A simple computational library to support water resources.
MIT License
2 stars 0 forks source link

could fda-model-cmd be a benchmark in fda-model-test? #187

Open Brennan1994 opened 2 years ago

Brennan1994 commented 2 years ago

Could we move the code in program to a test class in fda-model-test instead, removing this extra project? We could tag it with an attribute of [Catagory, Benchmark] to prevent it from running every pull request, and it would remove a project from our repo.

rnugent3 commented 2 years ago

Would that mean we need 6.0 to build fda-model-test?

Best,

R Nugent III

On Feb 25, 2022, at 9:26 AM, Brennan Beam @.***> wrote:

 Could we move the code in program to a test class in fda-model-test instead, removing this extra project? We could tag it with an attribute of [Catagory, Benchmark] to prevent it from running every pull request, and it would remove a project from our repo.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.

HenryGeorgist commented 2 years ago

.net 6.0 is a red herring here.

It would be fine to move it to be a test. The reason I wrote it as a command line application is that running an exe adds more features for profiling than simple tests. I can perform more profiling actions like evaluating thread usages and blockages which I needed to optimize the locking in our threadsafe histogram.

Brennan1994 commented 2 years ago

Gotcha. the implication being that we don't need those capabilities now because threadsafe histogram is already optimized enough for our application?

It won't hurt my feelings to leave the project. I'm just trying to do good faith organizational housekeeping.

HenryGeorgist commented 2 years ago

We should be testing for the purposes of benchmarking. Any change in any object may change the performance, we want to know when and why that happens. This project helps with that. If we want to not have 6.0 that is simple enough, just change the target, which will slightly change the main program class but no change in behavior.

rnugent3 commented 2 years ago

just to close the question of the version of .NET - I have no thoughts (or knowledge) about the one vs the other - I just wanted to know so I may begin requesting.

On Feb 25, 2022, at 6:22 PM, Will Lehman @.***> wrote:

 We should be testing for the purposes of benchmarking. Any change in any object may change the performance, we want to know when and why that happens. This project helps with that. If we want to not have 6.0 that is simple enough, just change the target, which will slightly change the main program class but no change in behavior.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.