# Cabal meeting (07/12/2023)

Previous meeting notes: https://hackmd.io/YmGTliI_Qf6kvlXwEzShLw

# Mandatory for the 3.10.3 release

Mikolaj:
- stuff to be backported are [here](https://github.com/haskell/cabal/issues?q=label%3A%22attention%3A+needs-backport+3.10%22) (each of the 4 ticket there at this time needs some extra work to finish/backport; volunteers welcome)
- we need a release 3.10.3 to not depend on an old (3.6) patched (with a security patch we already have in 3.10) GHCup version.
- no ETA, but ASAP
- GHC extensions, after a quickfix and the release, let's make a postmortem: why did we miss `TypeAbstractions`? (f-a will check)

# Fixing a serious pkg-config-related issue for the 3.12 release

Kristen:
- performance issue on `pkg-config` one
https://github.com/haskell/cabal/issues/8930
- Mikolaj: is a fix already on master? in 3.10? backportable? a cheap quickfix only for 3.10? --- the answers are "no"
- Mikolaj: Kristen, you are right, a related PR is merged to master, marked as needs-backport, but not backported yet: https://github.com/haskell/cabal/pull/9391
- Three ideas for the actual fixes are proposed:
- Caching all pkg-config queries
- Lazy IO for querying pkg-config only as needed
- Flag to disable pkg-config queries at start of dependency solving
- Two PRs for caching (not merged to master yet):
- https://github.com/haskell/cabal/pull/9360
- https://github.com/haskell/cabal/pull/9422

# `package-constraints` RFC
- [ticket](https://github.com/haskell/cabal/issues/9477)
- Javier illustrates them
- there will be a new section `package-contsraints`
- you can specify bounds there
- just syntactic sugar
- as now, have to topologically sort component in my mind,
which is error prone.
- questions: “should PC change _the solver_? If I depend on a package that depends on `aeson`, should that be transitive?”
Might be complicated, edge cases.
- name: `package-bounds` maybe? ?
- Michael: general constraints but `cabal check` say “don’t do it”
- because there are actually legitimate use cases, I will put them in the issue.
- it could be easier to explain to users.
- Gershom: syntactic sugar is convenient, transitive dependencies could be very weird for users.
- name: `default-package-bounds`?
- propose we stick to the initial syntax.
- Experimental release?
- brandon: it will be not used until it is accepted by Hackage, and then it will be frozen.
- Javier will summarise on the RFC ticket.

# Other stuff

Rodrigo:
* On applying for the second round of STF funding --- feedback is needed (please send to [e-mail]) on what to work on next, what to continue, what related work to tackle, e.g., one idea is per-module build plans that builds on the Hooks work
* Finished the re-design of coverage reporting https://github.com/haskell/cabal/pull/9464 (needs final reviews!)

Kristen:
* Status of https://github.com/haskell/cabal/pull/8999 -- @Hecate, do we need to redo the PR at our leisure later on?