Oceans of Wisdom

@[prxt0] / all, take a look and try, I think this is the right way to do it

Set up in GitHub. Created team `6529seize Maintainers` / `6529seize-maintainers` with repo permission `maintain` on `6529-Collections/6529seize-frontend`. Members active: `punk6529`, `ragnep`, `GelatoGenesis`, `simo6529`, `prxt6529` Effective repo roles verified: `punk6529`: `admin`, unchanged `GelatoGenesis`: `admin`, unchanged `ragnep`: `maintain` `simo6529`: `maintain` `prxt6529`: `maintain` Also replaced the old `main` branch protection with an active repository ruleset: `main maintainer review and merge policy` Ruleset ID: `18018081` It applies to the default branch and enforces: - PR required before merging - 1 approval required - approval must come from `6529seize Maintainers` - stale approvals dismissed on push - last pusher approval protection - review threads must be resolved - required checks preserved: `DCO`, `security/snyk (6529)` - no direct updates to `main` except via PR-only bypass - no deletes / force pushes The maintainer team has **PR-only bypass**, matching GitHub’s intended ruleset model: they can override and merge through a PR, while non-maintainers need one of them involved.

i think bot is amusing itself with E2E stuff

no, I don't think so

it can be in a separate PR

why not ask codex to fix the unrelated failures?

that's all

in a way that was simply not possible before

is that we can do mass scale improvements

and the general message I am here to preach

if the answer is a), then we need to have a different chat :slightly_smiling_face:

I am almost certain the answer is b)

what are the possible reasons that this was not done by you guys: a) you do not believe that testing is an important part of software development b) you thought it was going to be quite a bit of work and you were pressured by other priorities

i have no idea; have not looked at codex since this morning

the point that the whole team @[simo] @[GelatoGenesis] @[prxt0] should all take away as a learning was "why did you not harden and deepen and industrialize the testing framework before I came in and did it"

the point I am trying to make has nothing to do with this morning's PR

anyway, I am not actually upset at you to be clear

1/ last 60 hours completely rebuild the testing framework, a process that is I dunno 60-70% done 2/ @[prxt0] raises a PR during that process which fails a CI gate 3/ the PR is not, as far as I can tell, particularly urgent or important 4/ among the various available options: a) do not merge it yet b) fix the failing tests c) break my fairly complex in-progress rebuild of the testing framework, @[prxt0] suggest 4c

the correct series of events is the following

did I tell you to raise a PR before I finished doing the testing framework?

i mean if we are going to fully expand on it

we make similar decisions

so that when I am not here

the reason I am writing all this out is to explain my logic

and I am not saying this to just hear myself talk. I could just say "please prioritize this, not that"

so given all of this, obviously my POV is we should not kick the can down the road again and obviously not just on this PR, but on the whole 100x larger topics

3/ given #1 and #2, my #1 priority is getting this in place

2/ for this, we need extensivei, deep, broad, 'superhuman', 'unreasonable' amounts of testing

1/ the way to achieve all our hopes and dreams is to accelerate, parallelize, agent-ize etc

and so to wrap up this overly long text

for this we need tests, tests, more tests, additional tests, then more tests, followed by more tests and so on

the important point is that we have a system that we can trust to work autonomously

and of course even this is not the important point

and the amount of human time needed to write the prompt to fix 2 tests and to fix 1232435 tests is "exactly the same"

because the only scarce thing on this team is human time

is exactly the point of view I am trying to change

you see, the fact that you think that the number of failing tests matters, that if it was 2 tests, it was logical to fix them but if it is 1232435 tests, it was not logical to fix them

but on the 22nd of June, 2026, there is GPT 5.5 xhigh and Codex that CAN do this

and fix them while "i was doing something else altogether"

grind through

and you know, a year ago, there was no codex that could ground through 1232435 failing tests

the can would be kicked down the road again and again and again

and if I said "it is ok guys, no need to clean this up this morning"

yes because the tests and the repo have gotten out of line because at every point over the last years the team was not resolving this topic

whereas resolving the issue correctly took me a 5 minute prompt

I remember this from earlier times in human history like "this morning"

because you wanted to roll back the CI gate as opposed to fixing the PR

it took 2 hours and 45 minutes

but also I did prove my point on the tests :slightly_smiling_face:

not any AI