Only check with -Werror in CI

[locallycompact]

• Adds a nix flake check CI job to run on each PR which now also builds all components with -Werror enabled.

• Removes -Werror from cabal.project, to have the same behavior as nix build when building with cabal.

---

• CHANGELOG updated or not needed
• Documentation updated or not needed
• Haddocks updated or not needed
• No new TODOs introduced or explained herafter

[github-actions]

Transactions Costs

Sizes and execution budgets for Hydra protocol transactions. Note that unlisted parameters are currently using arbitrary values and results are not fully deterministic and comparable to previous runs.

Metadata
Generated at	2024-05-14 20:14:49.60912344 UTC
Max. memory units	14000000
Max. CPU units	10000000000
Max. tx size (kB)	16384

Script summary

Name	Hash	Size (Bytes)
νInitial	bccf2a430c016bc960fbf31b02694011cd399d20da8882aac9d33611	4110
νCommit	56b0f0b597150e619c76bed60683f3b1e42d7bc0685ed951b882bfc5	1975
νHead	86bff95ba20e9d1d1b34899a56d86bbacc9fed999260b27dcc92d128	9351
μHead	88f533cf67cd0fc93d7d9ccf0a8b1d69ffd1208a825efbebbc1d36ba*	4213

• The minting policy hash is only usable for comparison. As the script is parameterized, the actual script is unique per Head.

Cost of Init Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	4792	8.89	3.40	0.46
2	4997	10.85	4.15	0.49
3	5194	12.83	4.92	0.52
5	5603	16.77	6.43	0.58
10	6605	26.16	10.02	0.72
48	14244	99.97	38.37	1.86

Cost of Commit Transaction

This is using ada-only outputs for better comparability.

UTxO	Tx size	% max Mem	% max CPU	Min fee ₳
1	559	10.24	4.04	0.29
2	746	13.88	5.64	0.34
3	931	17.66	7.29	0.39
5	1312	25.66	10.74	0.49
10	2247	48.19	20.30	0.78
19	3935	97.83	40.79	1.41

Cost of CollectCom Transaction

Parties	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
1	56	543	16.88	6.66	0.36
2	113	654	26.81	10.68	0.47
3	171	764	38.32	15.41	0.60
4	228	874	52.99	21.40	0.77
5	282	984	67.62	27.51	0.94
6	338	1095	82.76	33.88	1.11

Cost of Close Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	549	7.89	4.26	0.27
2	688	8.59	5.32	0.29
3	906	10.16	7.06	0.32
5	1233	12.06	9.56	0.37
10	1913	15.12	14.71	0.46
50	7915	48.98	62.15	1.34

Cost of Contest Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	627	8.65	4.73	0.28
2	758	9.30	5.77	0.30
3	842	9.62	6.51	0.31
5	1147	11.39	8.94	0.36
10	2017	16.70	15.66	0.49
50	8002	51.88	63.42	1.38

Cost of Abort Transaction

Some variation because of random mixture of still initial and already committed outputs.

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	4669	17.00	7.35	0.55
2	4814	27.80	12.13	0.67
3	5052	43.43	19.16	0.86
4	5021	56.78	24.87	1.01
5	5088	70.57	30.80	1.17

Cost of FanOut Transaction

Involves spending head output and burning head tokens. Uses ada-only UTxO for better comparability.

Parties	UTxO	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
5	0	0	4627	8.15	3.40	0.44
5	1	57	4661	9.63	4.26	0.46
5	5	284	4796	13.70	6.88	0.52
5	10	569	4967	19.16	10.31	0.60
5	20	1140	5307	30.69	17.44	0.76
5	30	1707	5647	42.22	24.58	0.92
5	40	2278	5987	54.17	31.89	1.08
5	50	2847	6326	65.71	39.04	1.24
5	80	4552	7341	99.94	60.31	1.72

End-To-End Benchmark Results

This page is intended to collect the latest end-to-end benchmarks results produced by Hydra's Continuous Integration system from the latest master code.

Please take those results with a grain of salt as they are currently produced from very limited cloud VMs and not controlled hardware. Instead of focusing on the absolute results, the emphasis should be on relative results, eg. how the timings for a scenario evolve as the code changes.

Generated at 2024-05-14 20:17:12.053817995 UTC

Baseline Scenario

Number of nodes	1
Number of txs	3000
Avg. Confirmation Time (ms)	4.372424411
P99	10.252372399999985ms
P95	5.649074799999998ms
P50	4.1060585ms
Number of Invalid txs	0

Three local nodes

Number of nodes	3
Number of txs	9000
Avg. Confirmation Time (ms)	21.798442137
P99	51.69125948000007ms
P95	30.327334949999983ms
P50	19.6583045ms
Number of Invalid txs	0

[github-actions]

Test Results

427 tests  ±0   419 ✅ ±0   15m 32s ⏱️ -31s
139 suites ±0     8 💤 ±0 
  2 files   ±0     0 ❌ ±0 

Results for commit 23eae35. ± Comparison against base commit cf535c7.

♻️ This comment has been updated with latest results.

[ch1bo]

Why does this work? Why didn't it work before?

Our cabal.project does set -Werror on all local packages. So using cabal build I get the expected errors when doing:

rm -rf dist-newstyle && cabal build hydra-node

So why was the ci-nix.yaml workflow not failing on master? My hypothesis is that haskell.nix is not correctly picking up the project file and nix build .#hydra-node-tests does not fail (as does cabal build hydra-node above).

So my requested change: Clarify why it was not working before and fix it not only for nix flake check but for any builds through nix.

[locallycompact]

It's categorically a bad user experience to fail a nix run command on a branch because of a warning.

[ch1bo]

It's categorically a bad user experience to fail a nix run command on a branch because of a warning.

I don't have such a strong opinion and would be fine with allowing warnings in all our nix build invocations, while in CI we do add the -Werror.

However, my request would then be to make the behavior consistent with how cabal build works. Most likely this means removing -Werror from cabal.project and updating our CONTRIBUTING.md to reflect that change. I suspect we should say something along the lines of "while warnings are not treated as errors during builds, CI will check for it before we merge any contributions"

[ch1bo]

@locallycompact FYI I extended the PR description