- ghc-devs - Haskell.org

[CI] macOS builds
by Moritz Angermann 05 Jun '21

05 Jun '21

Hi there! You might have seen failed or stuck or pending darwin builds. Our CI builders we got generously donated have ~250GB of disk space (which should be absolutely adequat for what we do), and macOS BigSur does some odd reservation of 200GB in /System/Volumes/Data, this is despite automatic updates being disabled and time machine being disabled. It used to happen only when the system was expecting an update to be performed and the 200GB were freed after the update was done. After the latest update to 11.4, however, it seems to have not freed that space. This leaves the CI machine with ~50GB for for the system + build tools + gitlab checkouts and builds, and they frequently run out of space :-/ If someone knows how to prevent the system from doing stupid stuff like this (my hunch is it's keeping a backup of the system pre-udpate, for disaster recovery). Please come forward, my google searches haven't revealed anything useful yet. I have filed a TSI with Apple (still had a few on my developer account), but I don't expect them to come back to me before the end of June. Next week is WWDC, and there will be a massive backlog of issues that queued up leading up to, and during the WWDC. I've also only had very marginal success with them resolving issues that were not "you wrote this program wrong". If everything fails, maybe the solution is to attach some usbc ssd's to the macs and have gitlab builds be run dedicatedly on those disks. I'm a bit concerned about performance but we would have to see. Any ideas are welcome, please also feel free to hit me up on libera.chat#ghc, or the haskell foundations slack. Cheers, Moritz

1 0

value of documenting error messages?
by Richard Eisenberg 03 Jun '21

03 Jun '21

Hi devs, Take a quick look at https://gitlab.haskell.org/ghc/ghc/-/blob/6db8a0f76ec45d47060e28bb303e9eef6… <https://gitlab.haskell.org/ghc/ghc/-/blob/6db8a0f76ec45d47060e28bb303e9eef6…> You will see a datatype there with constructors describing error messages that GHC might produce. These constructors have comments describing the error, sometimes giving an example, and sometimes listing test cases. More datatypes like this one and more constructors in these datatypes are on the way. Question: Is there sufficient value in carefully documenting each constructor? In my ideal world, each constructor would have a high-level description, a detailed description of each field, an example of a program that generates the error, and one or more test cases that test the output. Along the way, we might discover that no such test case exists, and then we would add. However, generating this documentation is hard. I was thinking of whipping up an army of volunteers (Hécate has advised me how to do this) to do the work, but that army will need to be fed (with instructions, supervision, and reviews) and will want to know that their work is important. Is this effort worthwhile? Do we see ourselves maintaining this documentation? Or is the effort better spent elsewhere, perhaps tagging each constructor with an ID and then making wiki pages? Not sure what's best -- would love ideas! Credit to Alfredo di Napoli, who's done the heavy lifting of getting us this far. Relevant tickets: Original: https://gitlab.haskell.org/ghc/ghc/-/issues/18516 Tasks left: https://gitlab.haskell.org/ghc/ghc/-/issues/19905 Thanks, Richard

10 16

ghcup failed
by Simon Peyton Jones 02 Jun '21

02 Jun '21

Dear devs I wanted to install GHC 8.10 on my WSL2 (Windows Subsystem for Linux) computer. So I went here https://www.haskell.org/ghcup/ and followed the instructions (the curl … command). There was a long pause then [ Info ] verifying digest of: ghc-8.10.4-x86_64-deb9-linux.tar.xz [ Info ] Unpacking: ghc-8.10.4-x86_64-deb9-linux.tar.xz to /tmp/ghcup-khiegA [ Info ] Installing GHC (this may take a while) [ Error ] BuildFailed failed in dir "/tmp/ghcup-khiegA": NonZeroExit 2 "make" ["install"] Check the logs at "/home/simonpj/.ghcup/logs" and the build directory "/tmp/ghcup-khiegA" for more clues. Make sure to clean up "/tmp/ghcup-khiegA" afterwards. "_eghcup --cache install ghc recommended" failed! I looked in the logs as suggested, and found this in the tail of ghc-make.log Installing library in /home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/libiserv-8.10.4 "utils/ghc-cabal/dist-install/build/tmp/ghc-cabal-bindist" copy compiler stage2 "strip" '' '/home/simonpj/.ghcup/ghc/8.10.4' '/home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4' '/home/simonpj/.ghcup/ghc/8.10.4/share/doc/ghc-8.10.4/html/libraries' 'v p dyn' Installing library in /home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/ghc-8.10.4 "/home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/bin/ghc-pkg" --force --global-package-db "/home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/package.conf.d" update rts/dist/package.conf.install ghc-pkg: Couldn't open database /home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/package.conf.d for modification: {handle: /home/simonpj/.ghcup/ghc/8.10.4/lib/ghc-8.10.4/package.conf.d/package.cache.lock}: hLock: invalid argument (Invalid argument) ghc.mk:967: recipe for target 'install_packages' failed make[1]: *** [install_packages] Error 1 Makefile:51: recipe for target 'install' failed make: *** [install] Error 2 So I seem to be stuck. Any ideas? I feel embarrassed not to be able to install GHC 😊. Simon

4 4

GitLab downtime
by Ben Gamari 02 Jun '21

02 Jun '21

Hi all, Today I'm going to be working on migrating our GitLab instance to a new machine. Starting in likely an hour there will be a few hours of downtime while the migration proceeds. I will send an email when the downtime begins and concludes. Apologies for the late notice! Cheers, - Ben

6 10

Give MonadTrans a QuantifiedConstraints superclass
by Baldur Blöndal 01 Jun '21

01 Jun '21

This is to advertise the proposal (https://gitlab.haskell.org/ghc/ghc/-/issues/19922) to add a superclass to the MonadTrans type class in Control.Monad.Trans. A Monad transformer 'trans' lifts a 'Monad m' to a 'Monad (trans m)'. This proposal code-ifies that with a superclass constraint, an impliciation constraint enabled by the recent extension QuantifiedConstraints: class (forall m. Monad m => Monad (trans m)) => MonadTrans trans where .. This is the main motiviating example of the Quantified Class Constraints paper https://gkaracha.github.io/papers/quantcs.pdf

2 1

GHC releases and versions
by Sylvain Henry 28 May '21

28 May '21

Hi devs, We currently have 4 branches of GHC in flight: 8.10, 9.0, 9.2 and HEAD Latest releases: - 8.10.4: 2021/02/06 - 9.0.1: 2021/02/04 - 9.2.1-alpha2: 2021/04/23 Considering: 1) 8.10 branch should be stable but a lot of stuff has been merged for 8.10.5. To the point that 8.10.5 should probably be a "major release in the 8.10 series". 2) 9.0.1 is the latest major release but it shouldn't be used before 9.0.2 is released because of bugs and regressions (9.0.2 branch contains a fix for a critical bug in 9.0.1 [1] since March). 3) We might release 9.2.1 and 9.0.2 approximately at the same time which will be quite confusing for users ("9.0.2 in the 9.0 series and 9.2.1 in the 9.2 series"). 4) The first major number is meaningless. Proposition: Switch to A.B.C[.D] version scheme where: - A: major release ("series") - B: major release in the A series if B>0 and C=0; beta release if B=0 - C: bugfix release for A.B (if C>0) or beta release number (if B=0) - D: date when building in tree, not for releases It might be clearer exposed like this: showVersion = \case [a,b,c,d] -> "Dev version of " ++ showVersion [a,b,c] ++ " built on " ++ show d [a,0,c] -> "beta " ++ show c ++ " in series " ++ show a [a,b,0] -> "Major release " ++ show [a,b] ++ " in series " ++ show a [a,b,c] -> "Bugfix release " ++ show c ++ " for " ++ show [a,b] _ -> undefined > showVersion [9,0,1,20211028] "Dev version of beta 1 in series 9 built on 20211028" > showVersion [9,0,1] "beta 1 in series 9" > showVersion [9,0,2] "beta 2 in series 9" > showVersion [9,1,0] "Major release [9,1] in series 9" > showVersion [9,1,1] "Bugfix release 1 for [9,1]" > showVersion [9,2,0] "Major release [9,2] in series 9" > showVersion [10,1,0] "Major release [10,1] in series 10" Effects: 1) We could use C for bugfix versions which are to be released much faster than major versions. 2) B would be used for the old series we maintain. We backport a lot more stuff than before in older releases it seems, so it would be more PVP compliant to bump a major version number. 3) A would be used for the usual 6-month major releases. 4) We could make major releases in the 8 series (e.g. 8.10.5 could be released as 8.11.0) 5) We could advertise 9.0.1 as a beta (as everyone seems to consider .1 releases) 6) 9.2.1 final could be released either as 9.3 (next major in the 9 series if we just forget about 9.0.* and 9.2.*) or as 10.1.0 (first major in the 10 series). 7) No difference anymore between even/odd version numbers (for reference the current scheme is explained in [2]) Any thoughts? Sylvain [1] https://mail.haskell.org/pipermail/haskell-cafe/2021-March/133540.html [2] https://gitlab.haskell.org/ghc/ghc/-/wikis/working-conventions/releases

2 1

Errors in haddock git fetch
by Simon Peyton Jones 27 May '21

27 May '21

I'm getting these errors from `git submodule update`. Should I worry? Simon >From https://gitlab.haskell.org/ghc/haddock * [new branch] az/T19834 -> origin/az/T19834 * [new branch] az/T19834-2 -> origin/az/T19834-2 * [new branch] az/T19845 -> origin/az/T19845 * [new branch] az/T19845-2 -> origin/az/T19845-2 * [new branch] az/T19845-3 -> origin/az/T19845-3 * [new branch] dependabot/npm_and_yarn/haddock-api/resources/html/hosted-git-info-2.8.9 -> origin/dependabot/npm_and_yarn/haddock-api/resources/html/hosted-git-info-2.8.9 * [new branch] dependabot/npm_and_yarn/haddock-api/resources/html/lodash-4.17.21 -> origin/dependabot/npm_and_yarn/haddock-api/resources/html/lodash-4.17.21 * [new branch] dn/dn-driver-refactor-and-split -> origin/dn/dn-driver-refactor-and-split b4e7407b..c7281407 ghc-9.2 -> origin/ghc-9.2 dabdee14..4f9088e4 ghc-head -> origin/ghc-head * [new branch] wip/T18389-task-zero -> origin/wip/T18389-task-zero + 7d27ea7a...3b6a8774 wip/T19720 -> origin/wip/T19720 (forced update) + fe35fed3...40ba457f wip/adinapoli-align-ps-messages -> origin/wip/adinapoli-align-ps-messages (forced update) * [new branch] wip/dn-driver-refactor-and-split -> origin/wip/dn-driver-refactor-and-split error: cannot lock ref 'refs/remotes/origin/wip/hsyl20/dynflags': 'refs/remotes/origin/wip/hsyl20/dynflags/exception' exists; cannot create 'refs/remotes/origin/wip/hsyl20/dynflags' ! [new branch] wip/hsyl20/dynflags -> origin/wip/hsyl20/dynflags (unable to update local ref) * [new branch] wip/hsyl20/uncpp -> origin/wip/hsyl20/uncpp Unable to fetch in submodule path 'utils/haddock'; trying to directly fetch 4f9088e4b04e52ca510b55a78048c9230537e449: Submodule path 'utils/haddock': checked out '4f9088e4b04e52ca510b55a78048c9230537e449' simonpj@MSRC-3645512:~/code/HEAD-7$

3 2

Ampere A1 instances for builder/marge?
by Karel Gardas 26 May '21

26 May '21

Hi, just to pass https://www.servethehome.com/oracle-cloud-giving-away-ampere-arm-a1-instanc… -- oracle seems to be giving some "always free" Ampere/ARM based instances. Looks like they even have some RAM and cores so perhaps they may be usable for marge/builder for arm/linux platform? BTW: Not tested myself, not trust Oracle, but sometimes man can get some good even from ... well... Oracle... Thanks, Karel

1 0

Productivity Tip from mpickering
by Matthew Pickering 26 May '21

26 May '21

Hi, I am now using a new flavour combination which is proving *very* nice as a compromise between fast recompiles and passing tests. My normal build command is now: ./hadrian/build --flavour=default+no_profiled_libs+omit_pragmas --freeze1 -j This has the effect of * base libraries are compiled with optimisation * Profiling libraries are not built * Stage 1 compiler is compiled with -O + -fomit-interface-pragmas, so recompilation behaviour is much better. The end result is a nearly clean testsuite run (I think there are two failures) but much faster iterations when modifying `compiler/*`. Cheers, Matt

3 2

Coding style: Using StandaloneKindSignatures in GHC
by Baldur Blöndal 25 May '21

25 May '21

Discussion to permit use of StandaloneKindSignatures in the GHC coding style guide. I believe it increases the clarity of the code, especially as we move to fancier kinds. It is the only way we have for giving full signatures to type synonyms, type classes, type families and others. An example: type Cat :: Type -> Type type Cat ob = ob -> ob -> Type type Category :: forall ob. Cat ob -> Constraint class Category cat where id :: cat a a .. type Proxy :: forall k. k -> Type data Proxy a = Proxy type Some :: forall k. (k -> Type) -> Type data Some f where Some :: f ex -> Some f -- | The regular function type type (->) :: forall {rep1 :: RuntimeRep} {rep2 :: RuntimeRep}. TYPE1 rep1 -> TYPE rep2 -> Type type (->) = FUN 'Many This is in line with function definitions that are always given a top-level, standalone type signature (1) and not like we currently define type families/synonyms (2) by annotating each argument or not at all. Using -XStandaloneKindSignatures (3) matches (1) -- (1) curry :: ((a, b) -> c) -> (a -> b -> c) curry f x y = f (x, y) -- (2) type Curry (f :: (a, b) -> c) (x :: a) (y :: b) = f '(x, y) :: c -- (3) type Curry :: ((a, b) -> c) -> (a -> b -> c) type Curry f x y = f '(x, y) It covers an edgecase that `KindSignatures` don't. The only way for deriving to reference datatype arguments is if they are quantified by the declaration head -- `newtype Bin a ..`. StandaloneKindSignatures allows us to still provide a full signature. We could write `newtype Bin a :: Type -> Type` without it but not `newtype Bin :: Type -> Type -> Type` type Bin :: Type -> Type -> Type newtype Bin a b = Bin (a -> a -> b) deriving (Functor, Applicative) via (->) a `Compose` (->) a Let me know what you think

10 15