You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Right now, we use identifiers in our commits, issues, and PRs for the locations in the codebase that each of those affect. For example:
guest: add export_metadata macro
...is the message of a commit that touches hearth-guest. This is really clean and makes it really easy to compartmentalize what code changes, bugs, and ideas affect which specific parts of the repository.
However, now that Kindling is part of the codebase, what we've been doing so far is this:
kindling-panic-test: add export_metadata
Not as clean and not as readable. panic-test may also be a relatively short name for a service. Cutting the kindling- prefix out removes the context as to whether panic-test is a guest- or host-side crate.
We also need to consider that we have a plugin named hearth-init AND a Kindling crate named kindling-init, which implements the init system. Should one be renamed? Perhaps the hooks system that init implements could either be phased out or moved into hearth-runtime::utils.
Could we find an effective location scheme to use for Kindling crates that integrate into our existing location scheme? Should we reconsider this system altogether? Can we rely on understanding of the repository to provide context for commit messages and so on that do not have this information data?
I dislike Conventional Commits because it's always been my opinion that is much higher priority to describe the location of code changes than whether a code change is a bugfix or an addition.
This topic is not one that I can navigate alone. I'd like to discuss our options and form a consensus on the best way forward. Once we find one, we should describe it in the CONTRIBUTORS.md doc.
This is also really relevant to the topic of Kindling code sharing so I've included it in the "Kindling common code" milestone.
The text was updated successfully, but these errors were encountered:
I agree a lot that the location of code changes is really important and I think that us doing that has been one of the reasons that the commit log is so legible.
Perhaps we could have something like a single letter identifier to say whether its guest or host h-init: update flue version g-init: implement hot reload
Or just a K to mark it as kindling K-init: implement hot reload
Im not sold on this exact scheme but I think that a single letter could be better than a whole thing like kindling
K: panic-test: add export_metadata
The double colons here feel confusing.
I dont think that we should necessarily count on never having an overlap between host and guest crates. Our system should try to properly account for that.
As for longer service names, it may be a good idea to shorter them K-panel-manager: properly export caps K-panels: properly export caps
These would need to be well documented of course, we could do something similar to the main part of the repo where the folder name is the proper name for the commits but the actual crate name is longer and more specific.
Right now, we use identifiers in our commits, issues, and PRs for the locations in the codebase that each of those affect. For example:
...is the message of a commit that touches
hearth-guest
. This is really clean and makes it really easy to compartmentalize what code changes, bugs, and ideas affect which specific parts of the repository.However, now that Kindling is part of the codebase, what we've been doing so far is this:
Not as clean and not as readable.
panic-test
may also be a relatively short name for a service. Cutting thekindling-
prefix out removes the context as to whetherpanic-test
is a guest- or host-side crate.We also need to consider that we have a plugin named
hearth-init
AND a Kindling crate namedkindling-init
, which implements the init system. Should one be renamed? Perhaps the hooks system that init implements could either be phased out or moved intohearth-runtime::utils
.Could we find an effective location scheme to use for Kindling crates that integrate into our existing location scheme? Should we reconsider this system altogether? Can we rely on understanding of the repository to provide context for commit messages and so on that do not have this information data?
I dislike Conventional Commits because it's always been my opinion that is much higher priority to describe the location of code changes than whether a code change is a bugfix or an addition.
This topic is not one that I can navigate alone. I'd like to discuss our options and form a consensus on the best way forward. Once we find one, we should describe it in the CONTRIBUTORS.md doc.
This is also really relevant to the topic of Kindling code sharing so I've included it in the "Kindling common code" milestone.
The text was updated successfully, but these errors were encountered: