Useful categorisation

• I’d favour using colour to separate the different groups.
• I think we could collapse 4 and 5 into one group (one colour)

Simon

From: Ben Gamari <gitlab@gitlab.haskell.org>
Sent: 18 March 2019 21:39
To: Simon Peyton Jones <simonpj@microsoft.com>
Subject: GHC | Work out how to manage labels (#16458)

Currently ghc/ghc> has nearly 100 labels. These fall into a variety of categories:

1. Labels that classify the ticket (bug, feature request, task)
2. Labels that classify bugs by their manifestation: compiler perf, compiler crash, ghc build failure
3. Labels that specify the environmental conditions under which an issue will manifest:

• architectures: ARM, SPARC, PowerPC
• operating system: ~OpenBSD, Windows, macOS

4. Labels that specify which compiler or program features are implicated in a bug

• language extensions: ImpredicativeTypes, ~FlexibleContexts, UnboxedTuples
• program constructs: join points, levity polymorphism, STM
• compiler feature: recompilation checking, cpp, debug information, ~"cross compilation"

5. Labels that specify which bit of implementation a bug is likely caused by:

• compiler subsystem: driver, simplifier, demand analysis, RTS, LLVM backend
• related GHC project: ghc-pkg, integer-simple, nofib, core libraries, infrastructure
• external project: haddock, hadrian

6. Labels that specify other attributes of a ticket: newcomer, backport request, ~"info needed", ~"user facing"

It's not clear which of these are useful, which are just noise, and how we should best group them.

We have two tools for logically grouping labels:

• prefixing the label name with some sort of category string
• color

—
View it on GitLab.
You're receiving this email because of your account on gitlab.haskell.org. If you'd like to receive fewer emails, you can unsubscribe from this thread or adjust your notification settings.