-
-
Notifications
You must be signed in to change notification settings - Fork 13.9k
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
grpc: 1.47.0 -> 1.48.0 #182157
grpc: 1.47.0 -> 1.48.0 #182157
Conversation
@GrahamcOfBorg build bear sysdig libtensorflow |
@ofborg build google-cloud-cpp |
This broke some tests in |
I'm working on updating arrow-cpp to 9.0.0 and it looks like the tests pass with grpc 1.48.0. I'm still trying to work out some new dependency discovery failures, but we should be able avoid a revert here. |
First of all we should consider adding arrow-cpp to passthru tests of grpc if it easily breaks by updates and is an important consumer. Also the changelog does not mention any breaking changes I could find.
First of all: We are not going to start to hastily revert update because of single downstream failures. The fallout must be big/catastrophically or fixes require unjustifiable amounts of work on our side. If we can fix it by adding a patch from upstream then this is the preferred solution. In any way we should consider multiple solutions before reverting. If that means a package does not build for 5 days then so be it, that's okay and we will manage. master is not a branch where no (temporary) breakages are allowed to exist. When updating packages that are dependencies of many other packages then we must be careful if there are breaking changes. If we notice those early we should already include fixes in the update PR. If we notice those later we should do follow ups. If we immediately notice that the fallout is very big than a revert can be considered especially if the update is not very important. But it is not your responsibility to fix all (last percent(s)) downstream packages. Otherwise no one could or would want to maintain core packages because it also draws in partly maintenance of 100 or sometimes 1000 downstream packages. |
I think that's the right path moving forward. |
Core packages with this number of rebuilds should be merged to staging, not master. The lack of complete testing is compensated for by the opportunity to make fixes before breakage affects development of the >90% of PRs that target master. (packages with ~500 rebuilds live in an uncomfortable hinterland of course) |
Agreed, if a PR targets master it shouldn't break any downstream packages. That should be uncontroversial. |
reverse dependency is sensitive to changes as exhibited by NixOS#182157
Thankfully, it doesn't! |
How many reverse transitive dependencies does grpc have? 500 seems realistic to me, but I don't know the exact number |
What happens when a new version of gRPC breaks Arrow again, if #185329 is merged? What's the process to follow? gRPC would not be able to be updated until Arrow is updated? What if the Arrow update still doesn't work with that gRPC version? |
Sorry about that. I think #185124 is going to be merged in the next couple of days, so I'll wait that instead of reverting this PR. I maintain grpc because I use it in a hobby project, I'm not getting paid for it. |
Don't worry. We'll figure it out and the next time it breaks we just take a look what can be done to fix it. |
@marsam Please take care of yourself first. For better or for worse, Nixpkgs can't ensure that for you. |
@marsam Thanks for all the work that you do! I'm sorry to hear about your health, and best wishes for a swift recovery! |
Description of changes
https://github.com/grpc/grpc/releases/tag/v1.48.0
Things done
sandbox = true
set innix.conf
? (See Nix manual)nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD"
. Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/
)nixos/doc/manual/md-to-db.sh
to update generated release notes