Since renaming and removing labels can be easily done at any time after the migration, I decided to map most of the fields to labels during the migration, and we can then update the labels after the migration. Colors and descriptions can be updated too.
Merging label can technically be done after the migration too, by selecting all issues with label B, adding label A to all of them, and removing label B. Doing this after the migration however will generate two new events on each issue (addition of A and removal of B), and will update their âLast updatedâ date, making some searches more difficult. Because of this, doing it before the migration is better â the downside is that if we change our minds on a merge, itâs tricky to undo it.
Based on the feedback I received, I changed the following things:
Removed type-compile-error and type-performance and added a build and performance labels that can be combined with the other type-* labels
Temporarily renamed docs and tests to type-documentation and type-tests to match the existing labels in the CPython repo (these should be renamed)
Renamed type-enhancement to type-feature
Added labels for 3.7-3.11 (these can be removed afterwards)
Added release-blocker and deferred-blocker labels (these can also be removed)
Updated pending to map to pending instead of stale
Added labels for most components
Regarding the components, this is the full mapping:
Library (Lib)-> library
Documentation-> type-documentation
Interpreter Core-> interpreter-core
Windows-> OS-windows
Extension Modules-> extension-modules
Tests-> type-tests
asyncio-> expert-asyncio
IDLE-> expert-IDLE
Build-> build
email-> expert-email
IO-> expert-IO
macOS-> OS-mac
ctypes-> expert-ctypes
C API-> expert-C-API
Unicode-> expert-unicode
Installation-> expert-installation
Tkinter-> expert-tkinter
SSL-> expert-SSL
XML-> expert-XML
2to3 (2.x to 3.x conversion tool)-> expert-2to3
Cross-Build-> build (easily searchable, not too useful)
Demos and Tools-> `` (only a few issues, not too useful)
Subinterpreters-> expert-subinterpreters
Regular Expressions-> expert-regex
Argument Clinic-> expert-argument-clinic
FreeBSD-> `` (only a few issues, easily searchable)
Parser-> interpreter-core (only a few issues, easily searchable)
Distutils-> library (only a few issues, easily searchable)
FreeBSD and Demos and Tools have no corresponding labels, Cross-build and Build have been merged into build, Distutils has been included into library, Parser into interpreter-core.
This can be easily done either before or after the migration, and renaming type-behavior to type-bug is fine with me.
The python/cpython repo already has type-bugfix for PRs though, so there are a few options:
Rename type-bugfix to type-bug and use type-bug in the mapping, so that both issues and PRs will end up under the same label;
Get rid of the type-bugfix label for PRs and only mark issues with type-bug (issues are linked to PRs anyway, so we donât need to mark them twice unless some bot/script needs that label)
Keep both labels, use type-bug for issues and type-bugfix for PRs.
Iâve already renamed some of the labels on the python/cpython repo:
type-bugfix â type-bug (this will replace type-behavior too)
type-enhancement â type-feature
type-documentation â docs
type-performance â performance
type-tests â tests
I also noticed that the python/cpython already has some stage-like labels: awaiting change review, awaiting changes, awaiting core review, awaiting merge, awaiting review. bpo has these stages test needed, needs patch, patch review, commit review, backport needed, resolved.
Should we map patch review and commit review to awaiting review?
The awaiting * labels donât seem to be documented, so Iâm not entirely sure what mapping (if any) would make more sense.
The reason I was suggesting this mapping is that patch review/commit review are currently used on bpo to indicate issues that have a proposed solution (either a PR or a patch) ready to be reviewed. If we add the awaiting review, it will be possible to filter for these issues with is:issue is:open label:"awaiting review".
In theory this should be possible by searching for is:issue is:open linked:pr, but this has two problems:
PR linking is not supported during the migration (Iâm still looking for a solution)
Issues with a patch on bpo wonât be included in the search without a label
Moving forward it wonât be necessary to add this label to issues, since PRs will be linked properly, so itâs mostly a way to preserve some extra metadata of the migrated issues. If we decide against it, it will still be possible to search for issues awaiting for review on bpo even though the list will become outdated as more issues gets closed on GitHub.
commit review: A triager performed a patch review and it looks good. This signals to core developers the patch or pull request needs a quick once-over to make sure nothing was overlooked before committing it.
Either way there are only 21 issues with commit review, so it doesnât make much difference if they are not included. There are 2199 issues with patch review and also 2869 with the patch keyword (added automatically for PRs).
I actually want to filter them out. and searching GH seems like overkill. Why not just have a label and I can do what I want simply by adding that label (to the others that I currently use)?
Should we map 