-
Notifications
You must be signed in to change notification settings - Fork 121
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix benchmarking on travis (thanks to @tkf) #323
Conversation
Ok, I am maybe starting to understand what was going on. It looks like the problem in #322 was the
errors because the master branch already exists. What I don't understand is: why is it running on the master branch? I thought it was running on the branch of the PR? |
I think GitHub or Travis merges PR to the master branch before CI (I guess that's why the CI doesn't run when you have conflicts). So, maybe the solution could be to use |
Ok, that makes sense and is a clever solution; I just pushed a commit to try that. I don't quite understand why it works on the Transducers.jl repo though. Does the conditional |
Yeah, |
Ah ok, that makes sense. Looks like the benchmarks are running! Thanks for all your help :) |
Ok great, Travis works perfectly now. As an interesting aside, this PR (travis log link) gives us reliability info for Travis as a benchmarking platform, since there were no code changes (so everything should be invariant). Indeed, most of the benchmarks are invariant; the ones which aren't are: ––––––––––––––––––––––––––––––––––––––– ––––––––––– ––––––––––––
["ECOS", "GT_constraints"] 0.72 (5%) ✅ 1.00 (1%)
["ECOS", "LT_constraints"] 0.75 (5%) ✅ 1.00 (1%)
["ECOS", "equality_constraints"] 0.76 (5%) ✅ 1.00 (1%)
["SCS", "GT_constraints"] 0.77 (5%) ✅ 1.00 (1%)
["SCS", "LT_constraint"] 0.00 (5%) ✅ 0.00 (1%) ✅
["SCS", "LT_constraints"] 0.79 (5%) ✅ 1.00 (1%)
["SCS", "SDP_constraints"] 0.80 (5%) ✅ 1.00 (1%)
["SCS", "equality_constraints"] 0.77 (5%) ✅ 1.00 (1%)
["formulation", "GT_constraints"] 0.71 (5%) ✅ 1.00 (1%)
["formulation", "LT_constraints"] 0.72 (5%) ✅ 1.00 (1%)
["formulation", "equality_constraints"] 0.75 (5%) ✅ 1.00 (1%) All the benchmarks ending with So I think this might be a datapoint to suggest Travis is pretty decent for benchmarking low-allocation code but we should be wary of its reliability on high-allocation code. |
No description provided.