- ghc-devs - Haskell.org

GHC 9.8.3 plans
by Ben Gamari 30 Sep '24

30 Sep '24

Hi all, I am currently in the process of preparing our next release in the 9.8 series, 9.8.3. This release will include well over 70 backports. As I am a bit less than half-way through the backport process I have currently scheduled the release date for Wednesday 16 October 2024. I have included a list of currently-planned backports below (however it is possible that some of those in the to-do section may be dropped due to feasibility). If you know of anything else which you suspect should be backported do let me know on this thread and by applying the usual ~"backport needed:9.8" label. Cheers, - Ben # Already backported * [x] !12748: rts: fix checkClosure error message * [x] !12734: Various rts fixes for issues spotted by UndefinedBehaviorSanitizer * [x] !13252: Bump submodule array to 0.5.8.0 * [x] !13240: Bump submodule deepseq to 1.5.1.0 * [x] !13306: ci: Fix variable inheritence for ghcup-metadata testing job * [x] !13276: Fix interaction between fork and kqueue (#24672) * [x] !13151: JS: support rubbish static literals (#25177) * [x] !13113: AARCH64 linker: skip NONE relocations * [x] !13096: Cmm: don't perform unsound optimizations on 32-bit compiler hosts (#24893 #24700) * [x] !13091: GHCi debugger: drop record name spaces for Ids * [x] !13070: Only lookup ghcversion.h file in the RTS include-dirs by default. * [x] !13038: Address #25055, by disabling case-of-runRW# in Gentle phase * [x] !13010: Linker: Add windows R_X86_64_PC64 relocation support * [x] !12990: Fix -freg-graphs for FP and AARch64 NCG (#24941). * [x] !12885: PPC NCG: Fix sign hints in C calls * [x] !12975: X86 NCG: Fix argument promotion in foreign C calls * [x] !12914: Linker: use m32 allocator for sections when NEED_PLT (#24432) * [x] !12784: compiler: emit NaturallyAligned when element type & index type are the same width * [x] !12773: GHCi interpreter: Tag constructor closures when possible. * [x] !12713: user_guide: Fix typo in MultiWayIf chapter * [x] !12665: Add missing parenthesizePat in cvtp * [x] !12655: Reverse arguments to stgCallocBytes (fix #24828) * [x] !12574: doc: Fix type error in hs_try_putmvar example * [x] !12558: Add test cases for #24664 * [x] !12484: compiler: Fingerprint -fwrite-if-simplified-core * [x] !12463: JS: correctly handle RUBBISH literals (#24664) * [x] !12459: Don't depend on registerPackage function in Cabal * [x] !12432: Fix ghc API link in docs/index.html # To-do * [ ] !13237: Add `IsBootInterface` to finder cache keys * [ ] !12756: WIP: GHC.Cmm.Opt: Fix overflow on 32bit hosts (error) * [ ] !12639: Bump os-string submodule to 2.0.2.2 * [ ] !12607: refactor quadratic search in warnMissingHomeModules * [ ] !12603: Compatibility with 9.8.1 as boot compiler (error) * [ ] !12548: driver: always merge objects when possible * [ ] !12514: Track in-scope variables in ruleCheckProgram * [ ] !12500: GHCi: support cross-module inlining of breakpoints (#24712) * [ ] !12416: Don't generate wrappers for `type data` constructors with StrictData * [ ] !12370: NCG: Fix a bug where we errounously removed a required jump instruction. * [ ] !12344: Fix off by one error in seekBinNoExpand and seekBin * [ ] !12328: Improve the duplicate-tyvar check * [ ] !12321: Fix type of _get_osfhandle foreign import * [ ] !12313: driver: force merge objects when building dynamic objects * [ ] !12308: JS: fix for h$appendToHsString (#24495) * [ ] !12299: libffi-tarballs: bump libffi-tarballs submodule to libffi 3.4.6 * [ ] !12282: configure: Use LDFLAGS when trying linkers * [ ] !12264: rts: Lookup symbols in relevant DLLs only (originally !11164) * [ ] !12256: NCG: Fix a bug in jump shortcutting. * [ ] !12227: rts: fix clang compilation on aarch64 * [ ] !12184: Use "module" instead of "library" when applicable in base haddocks * [ ] !12177: Update correct counter in bumpTickyAllocd * [ ] !12162: driver: Make `checkHomeUnitsClosed` faster * [ ] !12161: Fix haddock source links and hyperlinked source * [ ] !12158: x86-ncg: Fix fma codegen when arguments are globals. * [ ] !12146: rts: avoid checking bdescr of value outside of Haskell heap * [ ] !12122: Improve base library synopsis and description * [ ] !12116: rel-eng/fetch-gitlab.py: Fix name of aarch64 alpine 3_18 release job * [ ] !12042: rts: only collect live words in nonmoving census when non-concurrent * [ ] !12016: Enhance documentation of Data.Complex * [ ] !12011: Add @since annotation to Data.Data.mkConstrTag * [ ] !11989: Fix ffi callbacks with >6 args and non-64bit args. * [ ] !11938: Escape multiple arguments in the settings file * [ ] !11575: configure: Do not override existing linker flags in FP_LD_NO_FIXUP_CHAINS * [ ] !11548: Update the unification count in wrapUnifierX * [ ] !11541: Compability with 9.8.1 as boot compiler (error) * [ ] !11384: hadrian: Move hsc2hs wrapper generation to bindist installation * [ ] !11383: testsuite: Run some tests unconditionally in nonmoving ways (error) * [ ] !11285: Draft: hadrian: Update bootstrap plans (error) * [ ] !11181: Bump Haddock to fix #23616 [GHC 9.8 branch] (error) * [ ] !10995: Fix (~) and (@) infix operators in TH splices (#23748) * [ ] !10890: nonmoving: Add support for heap profiling * [ ] !10659: nonmoving: introduce a family of dense allocators * [ ] !10523: Add missing BCO handling in scavenge_one (#23375) * [ ] !10417: StgToByteCode: Don't assume that data con workers are nullary (error) * [ ] !9762: nativeGen/AArch64: Fix graph-colouring register allocator

1 0

Gitlab access
by Vandenbroucke, Alexander 26 Sep '24

26 Sep '24

PUBLIC Hi ghc-devs, Can someone please approve my request for a gitlab account (alexander.vandenbroucke(a)sc.com<mailto:alexander.vandenbroucke@sc.com>). I would like to report a bug. Kind Regards, Alexander Vandenbroucke ---------------------------------------------------------------------- This email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please delete all copies and notify the sender immediately. You may wish to refer to the incorporation details of Standard Chartered PLC, Standard Chartered Bank and their subsidiaries together with Standard Chartered Bank’s Privacy Policy via our main Standard Chartered PLC (UK) website at sc. com

2 1

llvm 19 has been released
by George Colpitts 25 Sep '24

25 Sep '24

Hi llvm 19 has been released. Can we start using it in head? My smoke test of it with 9.10.1 was fine. Thanks George

4 5

New GitLab account approval
by Мария Мухина 22 Sep '24

22 Sep '24

Hello, I would like to get my GitLab account approved. My username is maria. Best regards, Maria

2 1

stack.yaml.lock gets dirty by building with build-stack
by ÉRDI Gergő 18 Aug '24

18 Aug '24

Hi, I have no local changes to hadrian/stack.yaml, but after running hadrian/build-stack, my hadrian/stack.yaml.lock file gets changed (all the custom package hashes are removed). Could someone check that the newly generated file is indeed the correct current one, and bump `master` if so? Thanks, Gergo

1 1

Can't initialise vendored submodule
by Arnaud Spiwack 18 Aug '24

18 Aug '24

Dear all, I'm currently unable to build GHC's master branch as some of the submodules (presumably used by Hadrian) are making git pretty unhappy (I've used various variants of the command below, in a cleaned repository): ```console $ git submodule update --init error: submodule git dir '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian/vendored/Cabal' is inside git dir '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian' fatal: refusing to create/use '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian/vendored/Cabal' in another submodule's git dir Failed to clone 'hadrian/vendored/Cabal'. Retry scheduled error: submodule git dir '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian/vendored/Cabal' is inside git dir '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian' fatal: refusing to create/use '/home/aspiwack/projects/tweag/ghc/master/.git/modules/hadrian/vendored/Cabal' in another submodule's git dir Failed to clone 'hadrian/vendored/Cabal' a second time, aborting ``` Please advise on how to get unblocked. And, if I'm allowed to conclude with a plea: please be careful not to abuse Git's submodule system, it's wonky enough when used simply. -- Arnaud Spiwack Director, Research at https://moduscreate.com and https://tweag.io.

4 4

HIE using ideclImplicit
by Brandon Allbery 14 Aug '24

14 Aug '24

I've been sitting on a potential proposal to enable per-import disabling of -Wunused-imports and -Wmissing-import-lists for a while now, and https://mail.haskell.org/pipermail/haskell-cafe/2024-August/136852.html has me thinking of moving forward with it. It turns out that this would be very easy to implement within ghc, as the ideclImplicit field of ideclExt (itself a field of ImportDecl) already does what's needed; it exists to track the implicit Prelude import, and would be almost trivial to set when parsing an import. I did a check to see if anything used it as a proxy for "this is the Prelude", and everything that wants that explicitly looks for the Prelude module, with one exception: HIE information excludes anything from an import with ideclImplicit set. My questions are: 1. Is this (omitting imports with ideclImplicit) even sane? What happens with HIE information when Prelude definitions are used in the module? 2. Is this use of ideclImplicit going to cause problems with my proposed use? 2a. Should HIE be checking explicitly for Prelude as other things do, or is it more correct for other imports with ideclImplicit set to be treated the same way? -- brandon s allbery kf8nh allbery.b(a)gmail.com

1 1

How to implement type-level list concatenation as a GHC primitive type family
by Hécate 10 Aug '24

10 Aug '24

Hi GHC devs, I was wondering recently, considering that type family evaluation is notoriously slow, how one would implement type-level list concatenation directly within GHC in a way that is much less expensive to run. So I am looking for pointers, as this part of GHC is fairly unknown to me. Thinking about it, I'm actually unsure where the tyfam evaluation is happening. Any advice would be appreciated. Cheers, Hécate -- Hécate ✨ 🐦: @TechnoEmpress IRC: Hecate WWW: https://glitchbra.in RUN: BSD

4 4

GHC 9.12 release schedule
by Zubin Duggal 07 Aug '24

07 Aug '24

Hi all, We are in the process of planning the GHC 9.12 release. The tentative schedule is a fork date in mid September with a final release in late November. If you are working on any major patches that should be included in this release, please be in touch. Furthermore, please ensure that all appropriate tickets or MRs are appropriately tagged with the %9.12.1 milestone (https://gitlab.haskell.org/ghc/ghc/-/milestones/393) You can also use the release tracking ticket for further discussion: https://gitlab.haskell.org/ghc/ghc/-/issues/25123 Cheers, Zubin

3 2

Reinstallable - base
by Simon Peyton Jones 07 Aug '24

07 Aug '24

Dear GHC devs Given the now-agreed split between ghc-internal and base <https://github.com/haskellfoundation/tech-proposals/pull/51>, what stands in the way of a "reinstallable base"? Specifically, suppose that - GHC 9.8 comes out with base-4.9 - The CLC decides to make some change to `base`, so we get base-4.10 - Then GHC 9.10 comes out with base-4.10 I think we'd all like it if someone could use GHC 9.10 to compile a library L that depends on base-4.9 and either L doesn't work at all with base-4.10, or L's dependency bounds have not yet been adjusted to allow base-4.10. We'd like to have a version of `base`, say `base-4.9.1` that has the exact same API as `base-4.9` but works with GHC 9.10. Today, GHC 9.10 comes with a specific version of base, *and you can't change it*. The original reason for that was, I recall, that GHC knows the precise place where (say) the type Int is declared, and it'll get very confused if that data type definition moves around. But now we have `ghc-internal`, all these "things that GHC magically knows" are in `ghc-internal`, not `base`. *Hence my question: what (now) stops us making `base` behave like any other library*? That would be a big step forward, because it would mean that a newer GHC could compile old libraries against their old dependencies. (Some changes would still be difficult. If, for example, we removed Monad and replaced it with classes Mo1 and Mo2, it might be hard to simulate the old `base` with a shim. But getting 99% of the way there would still be fantastic.) Simon

15 24