-
-
Notifications
You must be signed in to change notification settings - Fork 562
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
Bump depenencies (metrics) #2061
Conversation
No particular reason for it being pinned, likely part of past debugging Would you be able to fix the metrics error? What you describe as an issue with dependabot is actually something I quite like. I find it preferable to review each dependency at a time, and ensure that the changelog is properly reviewed - this is unlikely to happen with one giant change. Why is this required for Fedora? |
I mostly wanted to open the issue to show the errors. If you can fix them, please go ahead. I am not proficient in rust, I just monkey around in it, and throw stuff at it until the IDE shows me something useful.
Understandable. I personally prefer to get all dependencies resolved together. You still get each individual changelog when these are grouped, but you are also able to pick up changes that are dependent on each other. If you want to check individual changes, renovate would also allow you to do that by selecting which update you want in the PR.
Are you referring to how the changes of this PR are needed for Fedora?The relaxing of dependency is always useful, because otherwise we are making patches to do that on our side, which can be messy to maintain. In Fedora packaging, there is only one set of package available for each rust package, so there is no freedom like with Flatpak to use the exact dependencies specified in the project. So we often have to change the dependencies written there. If we see that the package still builds and tests pass with the current set of dependencies, than we go ahead with the packaging and wait for upstream/dependencies to catch up. But every so often there are breaking changes and we try to work together to make the necessary patches (there are people more rust proficient who get involved). This has the side-effect, that you get a bit more nudging from us whenever there are important dependency updates to bump 😉 |
I can work on fixing the metrics errors.
P.S.: I'll create a PR based off this one later this evening. |
Yeah the metrics dependabot PR is failing, so it's already visible. I'm more questioning whether updating that specific package is required for fedora or not?
Are you saying that there is only one version of each dependency that we can use for fedora? We're currently in several other distros and package managers with no such issues. I'm more than happy to make reasonable changes in order to make maintainers lives easier, but that doesn't sound particularly sustainable to me. What if we require a version of a dependency that fedora does not support? |
Sure, pls go ahead.
Yes, that's the case for most non-atomic distros. It may depend on the specific distro's guidelines, but thr general case is everything is shared and concise.
We always strive to update to the latest version as quickly as possible. If you update to a version newer than we have, we will follow suite and update them and the other dependants. If the dependencies are still compatible, which we check by the bundled tests, we silently make the changes. Otherwise we contribute patches upstream. Another model is that compatibility versions of dependencies are preserved. This however are rather special cases, and we instead strive to work with upstream first.
For packagers this is our bread and butter and we are used to it. We know that for a user the vendoring approach is more simple, but this model can be more beneficial for the whole ecosystem to spread updates and fixes, to promote active alternatives and so on. For the upstream developers side, you are not alone in this. I am primarily a C++/Python packager so I can't propose fixes as quickly, but usually we come with ready-fixes when it comes to bumping dependencies, although for tests it can be more complicated. I am pushing through packaging atuin because I really like using it and would like to make it more accessible, even if it means packaging 40+ dependencies. Once this is done though I will contribute some systemd units for the server package. |
Thanks for clarifying! Sorry for any potentially perceived pushback, but I found the approach to handling dependencies initially concerning, and just want to make sure I understand the implications of how you do things correctly So far, I've only worked with packagers that use either nix or rely on cargo for rust dependencies.
Thanks for this! Glad to hear you enjoy using it 😊 |
No worries at all, I've found the exchange very fruitful. I very much appreciate it when upstream gets involved in these topics. |
A quick dependency bump request to be compatible with the Fedora reviews
sqlx
dependency. Any reasons why it was pinned?metrics
dependencies bump. It seems the compilation now fails. I didn't look into the changelog to see how it should be fixedAbout dependabot, maybe you should consider
renovate
instead. One main issue with dependabot is that it continuously closes and opens PRs, even if you use the new group feature.renovate
on the other hand would always group and accumulate the dependency changes and rebase everytime on the same PR(s). This is helpful when you have multiple projects being closely linked to each other.Checks