- ghc-devs - Haskell.org

Re: [GHC] #4213: LLVM: Add support for TNTC to LLVM compiler suite
by David Terei 23 May '14

23 May '14

Firstly, its cool that LLVM has added this. It seems we could indeed implement TNTC with it. However based my quick understanding we couldnt implement it in a wag compatible with the current design. I believe the LLVM feature only allows us to place data at the very start of the function, after the label not before it. And your meant to ensure the very first instruction of the function is a jump that jumps over the data. We could use this, Simon Marlow and I discussed such a design in the past. But it's a change from the current scheme so the NCG would also need to change and it isn't clear if it would be as good as our current design. The other issue is that fixing TNTC doesn't eliminate the mangler. We use it for a few different reasons now. I can think of at least 3 I believe: TNTC, SIMD, and Windows. As for moving to the LLVM api. No, I think that isn't a great idea. I'm not convinced its will gain much compilation speed improvement. This is testable vy measuring how much time is actually spent invoking the LLVM binaries and how much time they spend serializing and parsing files. Given we need to go to an intermediary file of assembly for the mangler... Using the API also creates a GHC build time dependency on LLVM. Right now we avoid that which has advantages. On 22/05/2014 10:12 pm, "GHC" <ghc-devs(a)haskell.org> wrote: > > #4213: LLVM: Add support for TNTC to LLVM compiler suite > -------------------------------------+------------------------------------ > Reporter: dterei | Owner: dterei > Type: feature request | Status: new > Priority: low | Milestone: 7.6.2 > Component: Compiler (LLVM) | Version: 6.13 > Resolution: | Keywords: > Operating System: Unknown/Multiple | Architecture: Unknown/Multiple > Type of failure: None/Unknown | Difficulty: Unknown > Test Case: | Blocked By: > Blocking: | Related Tickets: > -------------------------------------+------------------------------------ > > Comment (by awson): > > I'm extremely sorry for a kind of OT. There is another LLVM/infotable > related ticket, which I don't know how to solve > [https://ghc.haskell.org/trac/ghc/ticket/8974 without the mangler]. Could > you, please, look into it and check if we can fix that without mangler or, > perhaps, understand what a feature we could ask LLVM people to introduce. > > -- > Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/4213#comment:14> > GHC <http://www.haskell.org/ghc/> > The Glasgow Haskell Compiler

1 0

Call for Contributions: Haskell Implementors' Workshop 2014
by Jost Berthold 21 May '14

21 May '14

Call for Contributions ACM SIGPLAN Haskell Implementors' Workshop http://haskell.org/haskellwiki/HaskellImplementorsWorkshop/2014 Gothenburg, Sweden, 06 September, 2014 Co-located with ICFP 2014 http://www.icfpconference.org/icfp2014/ Important dates --------------- Proposal Deadline: Monday, 14 July, 2014 (midnight, anywhere on earth) Notification: 21 July 2014 Workshop: Saturday, 6 September 2014 The 6th Haskell Implementors' Workshop is to be held alongside ICFP 2014 this year in Gothenburg. It is a forum for people involved in the design and development of Haskell implementations, tools, libraries, and supporting infrastructure, to share their work and discuss future directions and collaborations with others. Talks and/or demos are proposed by submitting an abstract, and selected by a small program committee. There will be no published proceedings; the workshop will be informal and interactive, with a flexible timetable and plenty of room for ad-hoc discussion, demos, and impromptu short talks. Attendance figures clearly reflect the growth of the Haskell user community and a constant interest in implementation aspects. Scope and target audience ------------------------- It is important to distinguish the Haskell Implementors' Workshop from the Haskell Symposium which is also co-located with ICFP 2014. The Haskell Symposium is for the publication of Haskell-related research. In contrast, the Haskell Implementors' Workshop will have no proceedings -- although we will aim to make talk videos, slides and presented data available with the consent of the speakers. In the Haskell Implementors' Workshop, we hope to study the underlying technology. We want to bring together anyone interested in the nitty-gritty details behind turning plain-text source code into a deployed product. Having said that, members of the wider Haskell community are more than welcome to attend the workshop -- we need your feedback to keep the Haskell ecosystem thriving. The scope covers any of the following topics. There may be some topics that people feel we've missed, so by all means submit a proposal even if it doesn't fit exactly into one of these buckets: * Compilation techniques * Language features and extensions * Type system implementation * Concurrency and parallelism: language design and implementation * Performance, optimisation and benchmarking * Virtual machines and run-time systems * Libraries and tools for development or deployment Talks ----- At this stage we would like to invite proposals from potential speakers for talks and demonstrations. We are aiming for 20 minute talks with 10 minutes for questions and changeovers. We want to hear from people writing compilers, tools, or libraries, people with cool ideas for directions in which we should take the platform, proposals for new features to be implemented, and half-baked crazy ideas. Please submit a talk title and abstract of no more than 300 words. Submissions should be made via EasyChair. The website is: https://www.easychair.org/conferences/?conf=hiw2014 If you don't have an account you can create one here: https://www.easychair.org/account/signup.cgi Because the submission is an abstract only, please click the "abstract only" button when you make your submission. There is no need to attach a separate file. We will also have a lightning talks session which will be organised on the day. These talks will be 5-10 minutes, depending on available time. Suggested topics for lightning talks are to present a single idea, a work-in-progress project, a problem to intrigue and perplex Haskell implementors, or simply to ask for feedback and collaborators. Organisers ---------- * Jost Berthold (co-chair) (University of Copenhagen) * Kevin Hammond (University of St.Andrews) * Gabriele Keller (University of New South Wales) * Paul Liu (Intel Labs) * Rita Loogen (Philipps-Universitat Marburg) * Geoffrey Mainland (co-chair) (Drexel University, Philadelphia) * Carter Schonwald (WellPosed Ltd) Contact ------- * hiw2014(a)easychair.org * Jost Berthold <berthold at di.ku.dk> * Geoffrey Mainland <mainland at cs.drexel.edu>

1 0

Questions about "time and space profiling for non-strict langs" paper and cost centre impl. of GHC
by Ömer Sinan Ağacan 21 May '14

21 May '14

Hi all, I'm trying to understand the paper "Time and space profiling for non-strict, higher-order functional languages"[1] and I'm hoping that experienced GHC devs here could help me clarifying some points. Here's my question: In Figure 5, I don't understand why no costs are attributed to `app2` rule(substitution rule). It looks like to me that this is the function application rule so `A` cost should be attributed here, but in the paper `A` cost is attributed in `app1` rule, which if I understand correctly shows how function part of an application is evaluated. I know about the "lexical scoping rule" explained in 2.4 and the cost should be attributed to the cost center in the scope of lambda definition instead of the cost center in "application site", but Figure 5 still doesn't make sense to me. To me it looks like that there should be two costs attributed in application rules. First cost is "evaluation of the function" part(which should be attributed in `app1` rule) and second is "substitution" part, which should be attributed in `app2` rule but according to the paper we only have the former cost, and the latter is free(e.g. no costs are attributed in second rule). It would be appreciated if someone could help me clarifying this. I'm also looking for more recent papers about GHC's profiler implementation. I'm especially interested profiling in multi-threaded RTS. Lastly, it'd be appreciated if someone could explain me two Bool fields in `StgSCC` constructor of STG syntax. Thanks! UPDATE: I was reading the paper one more time before sending my question and I found one more confusing part. In 4.1 it says "Lexical scoping will subsume all the costs of applying `and1` to its call sites..." but in 2.4 it says "Lexical scoping: Attribute the cost to "fun", the cost centre which lexically encloses the abstraction." so this two definitions of same thing is actually different. To me it looks like the definition in 4.1 is wrong,, it should be "definition site", not "call site". What am I missing here? [1]: http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=36FE097056F4E03CCD… --- Ömer Sinan Ağacan http://osa1.net

3 6

classP recently deleted from TH.Lib
by Gabor Greif 19 May '14

19 May '14

The last two commits from (Apr 9) on <https://github.com/ghc/packages-template-haskell/commits/master/Language/Ha…> removed a helper to construct applied type class constraints (`classP`). Some packages (notably `llvm-general-pure`) though, depend on it. Can we add back something like this: {{{ classP :: Name -> [Q Type] -> Q Pred classP cla tys = do tysl <- sequence tys return (foldl' AppT (ConT cla) tysl) }}} and export it again? Or is there such a helper with a different name already? Cheers, Gabor

7 10

Non-exhaustive pattern-match warning in code-example from "Dependently Typed Programming with Singletons"
by Herbert Valerio Riedel 19 May '14

19 May '14

Hello *, I've been experimenting with the code from the "Dependently Typed Programming with Singletons" paper[1] from the following is derived (modulo some irrelevant renamings): {-# LANGUAGE TypeOperators, DataKinds, GADTs, TypeFamilies #-} module CheckedList where data Nat = Z | S Nat data SNat n where SZ :: SNat Z SS :: SNat n -> SNat (S n) infixr 5 :- data List l t where Nil :: List Z t (:-) :: t -> List l t -> List (S l) t type family n1 :< n2 where m :< Z = False Z :< (S m) = True (S n) :< (S m) = n :< m index :: (i :< l) ~ True => List l t -> SNat i -> t index (x :- _) SZ = x index (_ :- xs) (SS i) = index xs i The problem, though, is that with the code above GHC 7.8.2 emits a warning for the `index` function: ,---- | Pattern match(es) are non-exhaustive | In an equation for ‘index’: Patterns not matched: Nil _ `---- So I would have expected to workaround that by explicitly wrapping the length-phantom with the promoted `S` type-constructor, like so index :: (i :< S l) ~ True => List (S l) t -> SNat i -> t index (x :- _) SZ = x index (_ :- xs) (SS i) = index xs i While this would probably have silenced the pattern-match warning, I now get a type-checking error I can't seem to get rid of: ,---- | Could not deduce (l1 ~ 'S l0) from the context ((i :< 'S l) ~ 'True) | bound by the type signature for | index :: (i :< 'S l) ~ 'True => List ('S l) t -> SNat i -> t | | or from ('S l ~ 'S l1) | bound by a pattern with constructor | :- :: forall t (l :: Nat). t -> List l t -> List ('S l) t, | in an equation for ‘index’ | | or from (i ~ 'S n) | bound by a pattern with constructor | SS :: forall (n :: Nat). SNat n -> SNat ('S n), | in an equation for ‘index’ | | ‘l1’ is a rigid type variable bound by | a pattern with constructor | :- :: forall t (l :: Nat). t -> List l t -> List ('S l) t, | in an equation for ‘index’ | | Expected type: List ('S l0) t | Actual type: List l1 t | Relevant bindings include | xs :: List l1 t | | In the first argument of ‘index’, namely ‘xs’ | In the expression: index xs i `---- Is there a way to tweak the code so that GHC doesn't think there's a non-exhaustive pattern-match? Cheers, HVR [1]: http://www.cis.upenn.edu/~eir/papers/2012/singletons/paper.pdf

3 3

RE: [commit: ghc] master: Extract derived constants from nm output for various OSes differently. (3df1c51)
by Simon Peyton Jones 19 May '14

19 May '14

Christian (and all ghc-devs) Thank you working so hard on this. I really appreciate it. But can I beg you to add a Note to explain the issue, and to refer directly to Trac #8783. The ticket has a string of forty-four comments, so presumably it's not a simple, obvious typo. What I fear is that in 5 yrs time someone will say "oh, there's a simpler way of doing this" and will un-do all your good work. Everyone: I'm know I get repetitive about this, but please always think about whether your change will seem as limpidly clear and obvious in five years time as it does now. ALWAYS add a reference to the ticket, unless you are simply fixing a totally stupid typo that nobody could possibly un-fix. (Stuff in the commit message is not enough, because you don't see it when you are looking at the source code. I enthusiastically encourage you to use the "Note" style, which you will see littered all over the compiler these days.) Pretty much any bug-fix is fixing something that wasn't obvious to the original author! Thank you all for making GHC so good. Simon | -----Original Message----- | From: ghc-commits [mailto:ghc-commits-bounces@haskell.org] On Behalf Of | git(a)git.haskell.org | Sent: 19 May 2014 06:11 | To: ghc-commits(a)haskell.org | Subject: [commit: ghc] master: Extract derived constants from nm output | for various OSes differently. (3df1c51) | | Repository : ssh://git@git.haskell.org/ghc | | On branch : master | Link : | http://ghc.haskell.org/trac/ghc/changeset/3df1c5109a6bd2a522717e524c10d8 | 42d4cd8ab8/ghc | | >--------------------------------------------------------------- | | commit 3df1c5109a6bd2a522717e524c10d842d4cd8ab8 | Author: Christian Maeder <Christian.Maeder(a)dfki.de> | Date: Thu Apr 3 10:00:07 2014 +0200 | | Extract derived constants from nm output for various OSes | differently. | | Fixes #8783. | | In order to avoid querying the nm version that does not work on Mac | OS X | we use the "nm -P" output that is supposed to produce (more | portable) | POSIX output and works on all tested OSes (MinGW, Mac OS X, Solaris | and | Linux using GNU nm) although slightly different (as documented). The | "nm | -P" output is actually only needed to recognize the output of a non- | GNU | Solaris nm (all other OSes produce sane outut using "nm" only). | | Signed-off-by: Austin Seipp <austin(a)well-typed.com> | | | >--------------------------------------------------------------- | | 3df1c5109a6bd2a522717e524c10d842d4cd8ab8 | utils/deriveConstants/DeriveConstants.hs | 37 +++++++++++++------------ | ------- | 1 file changed, 15 insertions(+), 22 deletions(-) | | diff --git a/utils/deriveConstants/DeriveConstants.hs | b/utils/deriveConstants/DeriveConstants.hs | index 8c943f0..6bfce24 100644 | --- a/utils/deriveConstants/DeriveConstants.hs | +++ b/utils/deriveConstants/DeriveConstants.hs | @@ -641,7 +641,7 @@ getWanted verbose tmpdir gccProgram gccFlags | nmProgram | oFile = tmpdir </> "tmp.o" | writeFile cFile cStuff | execute verbose gccProgram (gccFlags ++ ["-c", cFile, "-o", | oFile]) | - xs <- readProcess nmProgram [oFile] "" | + xs <- readProcess nmProgram ["-P", oFile] "" | let ls = lines xs | ms = map parseNmLine ls | m = Map.fromList $ catMaybes ms @@ -710,28 +710,21 @@ | getWanted verbose tmpdir gccProgram gccFlags nmProgram | doWanted (ClosurePayloadMacro {}) = [] | doWanted (FieldTypeGcptrMacro {}) = [] | | - -- parseNmLine parses nm output that looks like | - -- "0000000b C derivedConstantMAX_Vanilla_REG" | + -- parseNmLine parses "nm -P" output that looks like | + -- "derivedConstantMAX_Vanilla_REG C 0000000b 0000000b" (GNU | nm) | + -- "_derivedConstantMAX_Vanilla_REG C b 0" (Mac OS X) | + -- "_derivedConstantMAX_Vanilla_REG C 000000b" (MinGW) | + -- "derivedConstantMAX_Vanilla_REG D 1 b" | (Solaris) | -- and returns ("MAX_Vanilla_REG", 11) | - parseNmLine xs0 = case break (' ' ==) xs0 of | - (x1, ' ' : xs1) -> | - case break (' ' ==) xs1 of | - (x2, ' ' : x3) -> | - case readHex x1 of | - [(size, "")] -> | - case x2 of | - "C" -> | - let x3' = case x3 of | - '_' : rest -> | rest | - _ -> x3 | - in case stripPrefix prefix | x3' of | - Just name -> | - Just (name, size) | - _ -> Nothing | - _ -> Nothing | - _ -> Nothing | - _ -> Nothing | - _ -> Nothing | + parseNmLine line | + = case words line of | + ('_' : n) : "C" : s : _ -> mkP n s | + n : "C" : s : _ -> mkP n s | + [n, "D", _, s] -> mkP n s | + _ -> Nothing | + where mkP r s = case (stripPrefix prefix r, readHex s) of | + (Just name, [(size, "")]) -> Just (name, size) | + _ -> Nothing | | -- If an Int value is larger than 2^28 or smaller | -- than -2^28, then fail. | | _______________________________________________ | ghc-commits mailing list | ghc-commits(a)haskell.org | http://www.haskell.org/mailman/listinfo/ghc-commits

1 0

Status updates
by Austin Seipp 19 May '14

19 May '14

Hello all, It has been another slow week here, and there wasn't a lot I touched in the tree this week. The primary highlights: * The AMP battle rages on! It turns out the addition of `join` to `Monad` has had some subtle implications for GND/Roles. See here for more: https://www.haskell.org/pipermail/ghc-devs/2014-May/004964.html * Simon is still reviewing the ORF work, as far as I know. Hopefully it will Land Real Soon Now. * Patches! Lots of them! I'm reviewing and merging several of them right now. In particular, all the ones slated for 7.8.3 are up for grabs[1] by me right now, but I do need to double-check some of them e.g. on Windows, to make sure nothing breaks. * Many branches have been deleted. Something like 20-30 of them, I think. Yay! Sorry for the spam to the commits list too, but it was a necessary evil. * I'm working away at some of the bugs. #7602 is building right now, with some nice improvements. That will go into the 7.8.3 branch, so OS X users will get a big performance gain, yay! * Another bug: #8736. I had a patch for this, but this one is tricky to fix because when we load object files, we don't actually know if they're static or dynamic. This is encoded in the interface file, but there are some tricky semantics about what to do in some cases of (legitimately) needing to re-read interface files that are a bit subtle. I'll be trying to revive my patch for this today. Other than that, there really hasn't been a lot to mention. Please, do get down on reviewing some bugs and helping fix them![1] I'm more than willing to help in any way I can. I'd also like some feedback: what do people think about the current list? Are there any of you readers who see one *you* think is truly critical? Maybe we can get some input to help narrow the scope if needed. Thanks! [1] https://ghc.haskell.org/trac/ghc/query?status=infoneeded&status=merge&statu… -- Regards, Austin Seipp, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/

2 1

TypeLits vs. Z|Sn naturals
by Herbert Valerio Riedel 18 May '14

18 May '14

Hello again, ...while experimenting with TypeLits I stumbled over the following simple case failing to type-check with GHC 7.8.2: {-# LANGUAGE DataKinds, GADTs #-} data List l t where Nil :: List 0 t Cons :: { lstHead :: t, lstTail :: List l t } -> List (l+1) t with the error message ,---- | Could not deduce (l1 ~ l) from the context ((l + 1) ~ (l1 + 1)) | bound by a pattern with constructor | Cons :: forall t (l :: Nat). t -> List l t -> List (l + 1) t, | in an equation for ‘lstTail’ | | ‘l1’ is a rigid type variable bound by | a pattern with constructor | Cons :: forall t (l :: Nat). t -> List l t -> List (l + 1) t, | in an equation for ‘lstTail’ | | ‘l’ is a rigid type variable bound by | the type signature for lstTail :: List (l + 1) t -> List l t | | Expected type: List l t | Actual type: List l1 t | Relevant bindings include | lstTail :: List l1 t | lstTail :: List (l + 1) t -> List l t | | In the expression: lstTail | In an equation for ‘lstTail’: | lstTail (Cons {lstTail = lstTail}) = lstTail `---- While the code below happily type-checks: {-# LANGUAGE DataKinds, GADTs #-} data Nat = Z | S Nat data List l t where Nil :: List Z t Cons :: { lstHead :: t, lstTail :: List n t } -> List (S n) t Is this a known issue with the TypeLit solver? Cheers, hvr

3 2

Question about Roles and the AMP work
by Austin Seipp 15 May '14

15 May '14

Hello all, While picking the AMP patch back up, it is almost ready to go, but I got quite a puzzling error when attempting to build 'haskeline' with the AMP changes. I don't seem to remember encountering this error before - but maybe my mind is hazy. The error in question is this: https://gist.github.com/thoughtpolice/6df4c70d8a8fb711b282 The TL;DR is that the move of `join` into base has caused a problem here. The error really speaks for itself: GND tries to derive an instance of `join` for DumbTerm, which would have the type join :: DumbTerm m (DumbTerm m a) -> DumbTerm m a and be equivalent to join for StateT. But this can't hold: DumbTerm should unwrap directly to StateT (by newtype equality) and thus the third parameter to StateT in this case is `DumbTerm m a` if you expand the types - but because this argument is Nominal, it cannot unify with 'PosixT m', which is the actual definition of DumbTerm in the first place! With a little thought, this error actually seems reasonable. But it puzzles me as to what we should do now. One thing we could do is just write the instances manually, of course. But that's a bit annoying. Another alternative (and also unfortunate) change is to leave `join` out of Monad, in which case the GND mechanism is oblivious to it and does not need to worry about deriving these cases. I am not sure how many packages depend on using GND to derive monads like StateT with Nominal arguments, but I imagine it is not totally insignificant and will essentially hurt a lot of people who may use it. Richard, I'm particularly interested to hear what you think. This one sort of snuck up! (NB: despite this, the AMP work is very close to landing in HEAD, at which point we can begin cleaning stuff up faster.) -- Regards, Austin Seipp, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/

4 3

Many new validate failures, "Synonym in TyConAppCo"-related
by Joachim Breitner 13 May '14

13 May '14

Hi, I suddenly get Unexpected failures: gadt T7294 [exit code non-0] (normal) indexed-types/should_compile GADT1 [exit code non-0] (normal) indexed-types/should_compile GADT11 [exit code non-0] (normal) indexed-types/should_compile GADT3 [exit code non-0] (normal) indexed-types/should_compile IndTypesPerf [bad exit code] (normal) indexed-types/should_compile NonLinearLHS [exit code non-0] (normal) indexed-types/should_compile Numerals [exit code non-0] (normal) indexed-types/should_compile Roman1 [exit code non-0] (normal) indexed-types/should_compile Simple9 [exit code non-0] (normal) indexed-types/should_compile T2219 [exit code non-0] (normal) indexed-types/should_compile T2627 [exit code non-0] (normal) indexed-types/should_compile T3484 [exit code non-0] (normal) indexed-types/should_compile T4935 [exit code non-0] (normal) indexed-types/should_compile T7585 [exit code non-0] (normal) indexed-types/should_fail T2664a [exit code non-0] (normal) polykinds PolyKinds10 [exit code non-0] (normal) polykinds PolyKinds12 [exit code non-0] (normal) polykinds T6002 [exit code non-0] (normal) simplCore/should_compile T5329 [exit code non-0] (normal) simplCore/should_run T5315 [exit code non-0] (optasm) typecheck/should_compile T5490 [exit code non-0] (normal) typecheck/should_compile T7196 [exit code non-0] (normal) after these three changes were pushed: Changes to ghc: commit 214ad2d24cda4aaef541df3e213d5c4845f95c71 Author: Peter Trommler <ptrommler(a)scm.org> Date: Wed Apr 30 11:41:51 2014 +0200 Fix globalRegMaybe for unregisterised build. In commit 83a003f globalRegMaybe will be called but panics for unregisterised compilers. In an unregisterised compiler there are no global registers so always return `Nothing`. Fixes #9055. commit 21f17d06aa5c33e639f1b0d37b4bf888b494c441 Author: Simon Peyton Jones <simonpj(a)microsoft.com> Date: Tue May 13 13:17:19 2014 +0100 Fix invariant in mkAppCoFlexible mkAppCoFlexible was breaking the invariant that the head of a TyConAppCo cannot be a type synonym. This small patch fixes it. commit 4cfc1fae11ec9a5c4b34ac747f0ce50f52423eba Author: Simon Peyton Jones <simonpj(a)microsoft.com> Date: Tue May 13 13:15:45 2014 +0100 Lint should check that TyConAppCo doesn't have a synonym in the tycon position That is why Lint didn't nail Trac #9102 Full log at https://s3.amazonaws.com/archive.travis-ci.org/jobs/25063931/log.txt. It seems that most, if not all, fail with a "Synonym in TyConAppCo" lint error. If Lint is correct here, then this may indicate further bugs are around. But in that case the correct thing would be to have a ticket number for that problem and mark all failing test cases as expect_broken – it’s the only sensible way of having an always succeeding test suite in master. So is the Lint check correct and this uncovered one or more bugs here, or is the Lint check actually too restrictive? Greetings, Joachim Greetings, Joachim -- Joachim “nomeata” Breitner mail(a)joachim-breitner.de • http://www.joachim-breitner.de/ Jabber: nomeata(a)joachim-breitner.de • GPG-Key: 0xF0FBF51F Debian Developer: nomeata(a)debian.org

4 3