Closed mordamax closed 1 year ago
I would like to rewrite this project and prepare it for the Fellowship migration.
The current state of the code is too messy, there is no abstraction at all which makes this so difficult to implement.
And the configuration files are not readable at all:
exclude: ^\.gitlab-ci\.yml|^scripts/ci/.*|^\.github/.*|^\.config/nextest.toml|^frame/(?!.*(nfts/.*|uniques/.*|babe/.*|grandpa/.*|beefy|merkle-mountain-range/.*|contracts/.*|election|nomination-pools/.*|staking/.*|aura/.*))
include: ^frame/(?!.*(nfts/.*|uniques/.*|babe/.*|grandpa/.*|beefy|merkle-mountain-range/.*|contracts/.*|election|nomination-pools/.*|staking/.*|aura/.*))
include: ^\.gitlab-ci\.yml|^scripts/ci/.*|^\.github/.*|^\.config/nextest.toml
exclude: ^runtime/(kusama|polkadot)/src/[^/]+\.rs$|^\.gitlab-ci\.yml|^(?!.*\.dic$|.*spellcheck\.toml$)scripts/ci/.*|^\.github/.*
So, I have been researching the code:
Most rules here and AND DISTINCT
rule here set the reviews here.
We need to add an extra review by the user itself and let the system check if it counts.
Parity code should in general require 3 Parity people to know it prior to merge. This includes the author, therefore we generally want 2 reviews.
@gavofyork do I understand correctly, considering that we will start to count PR initiator in according team condition, that we should also increase min_approvals to 3 for core-devs condition?
a) Let's say one of core-devs created PR in:
non-locked place:
locked place:
b) Non-core-dev creates PR in:
Reopened for now, until we test the fixes
Fixes had been tested.
Current staging server instance is running with the new update.
As me and @mordamax are away until Friday, we decided to postpone the production release as we won't be able to monitor it during our absence.
@gavofyork do I understand correctly, considering that we will start to count PR initiator in according team condition, that we should also increase min_approvals to 3 for core-devs condition?
As I understood the request, people that are being part of "special groups" (aka not core-devs
) count towards the "approval" count. So, we always require two core-devs
and the author is not counting towards these minimum number of two. However, when the pr also requires lock-review
and being part of this group, it counts towards these two.
@bkchr As I understood Any parts of the codebase may also have their own lists of reviewers. These can continue, but individual reviews are allowed to contribute to multiple sub-conditions. It should still be based around "knowers-of-code", which means the author should count towards the requirements.
any part, including for core-devs
and Parity code should in general require 3 Parity people to know it prior to merge. This includes the author, therefore we generally want 2 reviews.
to me would mean that regardless of lock-review
the "knowers-of-code" is always the rule (which makes sense).
min_approvals to 3 for core-devs condition
If not "knower" would contribute - that would be these 3 Parity people to know it
if "knower" -> then generally want 2 reviews
would be, discounting the "knower" themself
just wanted to confirm
But then an external contributor will require three reviews. Not sure that we want this. core-devs
is some special group and we should just "ignore", IMO.
@bkchr, yes, and it seems to be corresponding to this requirement
Parity code should in general require 3 Parity people to know it prior to merge
I'm curious, whether it's more often happen that core-devs open PRs daily or external contributors? To what i see in PR list - it's more often core-dev authors, so in most cases this change is unnoticeable for people in parity. Thinking what kind of a problem for external devs (or us) that is going to become if +1 core-dev (out of 70+) would have to approve this?
Just because people are part of core-devs
, doesn't mean that they have any more knowledge than an external contributor. But yeah, let's try it. If not we can revisit.
Ok, i've merged bump to 3 core-devs in polkadot. so this should be closed now
May 25
Upd Jun 18: