November 2015 - ghc-devs

Pre-Proposal: Introspective Template Haskell
by Richard Eisenberg 14 Nov '15

14 Nov '15

Hi devs, There's a bunch of open tickets around Template Haskell. A great many of them are attempts to make TH more like what's already in GHC. Of course, when someone improves GHC, then TH also has to be updated. But this doesn't always happen, leading to several of these tickets. I believe I have a solution to the problem: just eliminate Template Haskell and provide direct access to GHC's internal structures. The idea (still very sketchy; hence pre-proposal) is like this (numbered for easy reference, but no order is implied): 1. TH quotes would remain. DsMeta would desugar quotes into Core code that produces HsExprs. For example, [| 1 |] would have type (Q (LHsExpr Name)). (Or perhaps (Q (LHsExpr RdrName)) if that works out better for clients.) 2. TH splices would remain, working much as they do now. The expression inside, say, an expression splice would have type (Q exp) where we can satisfy the constraint (SpliceExpr exp). There would be instances for (SpliceExpr (LHsExpr Name)) and (SpliceExpr (LHsExpr RdrName)) as well as the non-located variants. Generalizing the type of expressions here allows users not to worry about un-renaming when roundtripping between quotes and splices. 3. Reification would remain, using an Info structure much like we have now. Would we expose the real GHC TyCons as the result of reification? Or is it better to give the users HsDecls? This would need to be fleshed out. 4. Lifting would remain, doing the obvious thing. 5. The template-haskell package could even remain, as a backward-compatibility shim. It would declare gobs of pattern synonyms and type synonyms to wrap the new underlying interface. This re-imagined template-haskell package would not need to be a boot library, and could be upgraded separately from GHC. We could even maintain multiple versions of the library so that TH clients wouldn't have to change their code when GHC upgrades. Perhaps someday we could think about deprecating, if that's the way the wind blows. So, the end result is a completely re-engineered TH, but I believe we could keep full backward compatibility. (I have not considered Typed TH in any depth yet. But my hope is that it's not too different from the above.) And, we're left with a significantly lower maintenance burden, especially if we eliminate template-haskell someday. And, tantalizingly, the flexibility in splices might allow us to splice in *Core* code someday. Perhaps we could also reify Core code. Then clients could write their own custom, domain-aware optimizations. Like RULES on steroids. But that's all for the next phase. (Giving due credit, this last bit is inspired by work David Christiansen is doing in Idris.) What's wrong with this idea? I feel like *something* has to be more complicated than I've made it seem! Richard

6 15

On the record pattern synonyms patch
by Edward Z. Yang 13 Nov '15

13 Nov '15

Hello Matthew, I have a question about the record pattern synonyms patch which you landed in the last month (2a74a64e8329ab9e0c74bec47198cb492d25affb). In IfaceSyn, you made the following change: data IfaceIdDetails = IfVanillaId - | IfRecSelId IfaceTyCon Bool + | IfRecSelId (Either IfaceTyCon IfaceDecl) Bool | IfDFunId I don't understand why IfRecSelIds can now store IfaceDecls. Why aren't these declarations stored at the *top level* list of declarations in an interface file? If they are, you shouldn't be storing the entire IfaceDecl structure when serializing to disk. If they are not, I am afraid I don't understand this code. I also noticed that the commit message is quite long, and many of the comments I see are not replicated anywhere in the source. Is there any reason why they should not be written down as proper notes, with cross-references from the appropriate code? Thanks, Edward

2 1

Haskell Platform Plans
by Gershom B 13 Nov '15

13 Nov '15

Dear all, As you know, Mark has passed on the Haskell Platform maintainer hat to me. I wanted to give a heads-up on current plans to keep folks in the loop. This message has three sections, first the 7.10.3 release, next the 8.0 release, and finally some general musings on fiddly details and future plans. 1) 7.10.3 The 7.10.3 release candidates have been out for windows and unix for a bit. As I understand it there is still work underway on the mac build, but that will hopefully be coming shortly. The plan is to release a new platform roughly simultaneously. This platform will work essentially as in the past, for two reasons. First, because it is the last release in the 7.10 series and should be seen like the 7.10.3 compiler as just incorporating some necessary patches rather than any serious changes. Furthermore, the future plans underway rely on a few patches to cabal which have been merged, but are not yet in any existing cabal-install release. A few packages have received minor version bumps, as follows: PACKAGE 7.10.2-a latest ------------------------------------------- case-insensitive 1.2.0.4 1.2.0.5 fgl 5.5.2.0 5.5.2.3 GLUT 2.7.0.1 2.7.0.3 GLURaw 1.5.0.1 1.5.0.2 HUnit 1.2.5.2 1.3.0.0 OpenGL 2.12.0.1 2.13.1.0 OpenGLRaw 2.5.1.0 2.6.0.0 primitive 0.6 0.6.1.0 syb 0.5.1 0.6 scientific 0.3.3.8 0.3.4.2 StateVar 1.1.0.0 1.1.0.1 A few packages were held back due to dependency issues (notably zlib) and additionally, packages shipped with ghc were held back to the versions that will be shipped with 7.10.3. 2) 8.0 We also plan to ship a new platform concurrently with the ghc 8.0 that should be released early next year. That platform will implement some long-discussed and requested changes. First, the platform already ships with msys on windows. However, it does not do so in a way that lets you, as with minGHC, install the network library directly, or for that matter install directly other libraries that use build-type configure. This is because the msys binaries don't get placed into the path, by design. The new platform will ship with a newer cabal, incorporating a patch (pull request #2480) that uses the extra-prog-path setting for build-type: configure. This should allow network and other things reliant on build-type: configure to directly cabal install. The goal for the platform on windows will be that any packages binding to msys libraries _should_ be cabal installable without hassle. If you maintain a library that you think would be a good test-case for this, please drop a line so we can work together towards this goal. Second, the default platform will be _minimal_. That is to say that it will _only_ install into the global package database those packages that are directly shipped with ghc, and not any additional platform packages. "Blessed" platform packages will however still be shipped as a "default user cabal.config" file, so in a setting where users want to work unsandboxed, their versions of platform libs will remain pegged to those chosen by the platform, should they not choose to alter their settings. In a sandboxed setting, or in the presence of a local cabal.config generated by a freeze, those constraints will be disregarded. Furthermore, it is likely but not certain that the "nix-like cabal" or "no-reinstall cabal" will be available for the 8.0 release. If this is the case, users will be able to operate in an unsandboxed environment without conflicts, as multiple instances of the same version of a package, with different dependency choices, will be able to live side-by-side. Third, the platform will ship not only with cabal-install, but also with the stack tool, so that users can choose either workflow, depending on their project or preferences. A number of people have adopted the stack tool, and many beginners have reported success with it as well, so it will be good to provide that functionality out of the box with every platform install. Time and resources permitting we would also like to ship a platform version _with_ the platform libraries prebuilt and included, as that is also a use case that some people continue to like. However, this is a secondary priority, and contingent on us getting our build infrastructure into a good enough shape to make this not too much extra hassle. I'm excited about these changes, and confident we can get their in a timespan that matches the 8.0 release of ghc. 3) Generalities As I mentioned, it would be good to have a more uniform build infrastructure. Generally, I have put out some feelers and am working towards extending the ghc nightly buildbot infrastructure to both cover more platforms and also cover more tools -- not only ghc, but cabal, the platform, perhaps haddock, ghcjs, etc. This way we can get an economy of scale and many of our core tools will be able to all share regular automated builds across many platforms. If you like CI/buildbot systems and would like to help out with this project, please reach out. Also, once we've modernized and fixed up the core platform installer tech, it would be nice to move back to the process of making the platform a good set of blessed libraries -- taking more proposals for additions to it, looking to cull libraries that are no longer widely used, etc. As part of this I intend to move the haskell-platform list off our deprecated community.haskell.org infrastructure and onto mail.haskell.org with our other active lists. Finally, I'm happy to be maintainer of the platform through this period of change and transition, but at some future point, as things get sorted out and the release process becomes more standard and mechanical, I would very much like to pass this responsibility on. I have had some nibbles of offers, but if other people want to begin to get involved, please let me know and we can start to get small contributions from you so that you can become familiar with the various tech and systems involved. The Haskell ecosystem is a team effort, a collective project, and volunteer driven. In my modest experience hacking on the various systems involved (cabal, cabal-install, hackage, the platform build, etc.) some are a bit confusing, but I have always found helpful contributors willing to explain things and review patches, and to help think about and diagnose problems. And none of the code has been as confusing as things I've had to wade through for employment-related purposes :-). So when this stuff doesn't work as well as we'd like, we can investigate together, and then put our heads together and figure out how to improve it together. Furthermore, it can be very satisfying to do so, because the impact of those improvements makes itself widely felt. I look forward to more people joining in! Best, Gershom

1 0

Re: [GHC] #11018: ApiAnnotations: Add SourceText for unicode tokens
by Alan & Kim Zimmerman 13 Nov '15

13 Nov '15

agree, as we discussed on IRC On Thu, Nov 12, 2015 at 9:30 PM, GHC <ghc-devs(a)haskell.org> wrote: > #11018: ApiAnnotations: Add SourceText for unicode tokens > -------------------------------------+------------------------------------- > Reporter: alanz | Owner: alanz > Type: bug | Status: new > Priority: normal | Milestone: 8.0.1 > Component: Compiler | Version: 7.10.2 > Resolution: | Keywords: > | ApiAnnotations > Operating System: Unknown/Multiple | Architecture: > | Unknown/Multiple > Type of failure: None/Unknown | Test Case: > Blocked By: | Blocking: > Related Tickets: | Differential Rev(s): > Wiki Page: | > -------------------------------------+------------------------------------- > > Comment (by mpickering): > > No. Go for it. > > For the purpose of this ticket, I will also add that I think adding a flag > `IsUnicodeSyntax` is better than adding `SourceText`. > > -- > Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11018#comment:2> > GHC <http://www.haskell.org/ghc/> > The Glasgow Haskell Compiler >

1 0

Questions on the RTS C API regarding threads and tasks
by Nicola Gigante 12 Nov '15

12 Nov '15

Hi everybody, I’m new on this list, congratulations to everyone for the great work that you’re doing on GHC! I apologize if this is not the right list where to ask the following questions. Please point me to the correct list or IRC channel otherwise. I’m part of a small group of people who is going to work on a project that seems to need low-level handling of haskell threads. I’ve started delving into the ghc runtime API to understand if everything we need is exposed to the world or if we have to modify the runtime itself (which I would avoid if possible). I have a few questions about the functions that the runtime exports to the C world regarding the manipulation of tasks: - Fundamentally, is there a way for a C function called by a foreign call to suspend the haskell thread that called it, to be resumed later when appropriate? I see that the runtime has a concept of “blocking queue”, but I also see that the functions that work on blocking queues are not exported. I can manage a queue of TSOs myself, but then I need a way to put in sleep the task indefinitely up to a wake signal. In other words, can I sleep and awake haskell threads from a C function called by a foreign call? - I see the “suspendThread” and “resumeThread” functions exist in include/rts/Threads.h Are these the functions that I need for the point above? My main concern is that the comments in the source code for suspendThread say that these functions are used in case of a CCall, and put the thread in a specific queue, so they seems to have more specific role than what the name might suggest. - Also, I’m unable to find what a StgRegTable is (which is taken as argument by suspendThread). Thank you in advance for your help, Greetings, Nicola

3 3

one weird trick to get all your types
by Richard Eisenberg 12 Nov '15

12 Nov '15

Hi devs, I spent some of my time at Hac Phi this past weekend teaching a few people the ropes about GHC hacking. One innocently asked if it's possible to load GHC into GHCi. I shuddered at the thought. But, actually, you can get quite close to this ideal: > ghc-stage2 --interactive GHCi, version 7.11.20151104: http://www.haskell.org/ghc/ :? for help Prelude> :set -package ghc -- THIS IS THE KEY LINE package flags have changed, resetting and loading new packages... Prelude> import TcMType Prelude TcMType> :t newEvVar newEvVar :: forall gbl lcl. TcType.TcPredType -> TcRnTypes.TcRnIf gbl lcl Var.EvVar Prelude TcMType> Pretty cool, huh? This improves my life. Hopefully it improves yours, too. Richard PS: Can anyone recommend a place on the wiki where I should put this?

2 1

Pretty-printing TH data types
by Jan Stolarek 11 Nov '15

11 Nov '15

Devs, Module Language.Haskell.TH.Ppr defines Ppr instances for Template Haskell data types. What is the fastest way of testing these instances, ie. pretty-printing TH data types? I want to make sure that the instances I added print the data type in the way that I expect. Janek --- Politechnika Łódzka Lodz University of Technology Treść tej wiadomości zawiera informacje przeznaczone tylko dla adresata. Jeżeli nie jesteście Państwo jej adresatem, bądź otrzymaliście ją przez pomyłkę prosimy o powiadomienie o tym nadawcy oraz trwałe jej usunięcie. This email contains information intended solely for the use of the individual to whom it is addressed. If you are not the intended recipient or if you have received this message in error, please notify the sender and delete it from your system.

2 2

How inline pragma works
by Эдгар Жаворонков 11 Nov '15

11 Nov '15

Hello everyone! Where can i read about how does INLINE pragma in compiler works? --- С уважением, Жаворонков Эдгар Best regards, Edgar A. Zhavoronkov

2 3

Problems with cabal
by Эдгар Жаворонков 10 Nov '15

10 Nov '15

Hi, everyone! I have a strange problems with building GHC from sources. I cloned git repo as said in newcomers' page, but i have that kind of error: Cloning into 'libraries/Cabal'... fatal: repository 'https://github.com/ghc/packages/Cabal.git/' not found Clone of 'http://github.com/ghc/packages/Cabal.git' into submodule path 'libraries/Cabal' failed What am i doing wrong? --- С уважением, Жаворонков Эдгар Best regards, Edgar A. Zhavoronkov

2 3

Re: Questions on the RTS C API regarding threads and tasks
by Nicola Gigante 10 Nov '15

10 Nov '15

Il giorno 05 nov 2015, alle ore 05:27, Edward Z. Yang <ezyang(a)mit.edu> ha scritto: > > Excerpts from Nicola Gigante's message of 2015-11-04 11:12:39 -0800: >> I’ve started delving into the ghc runtime API to understand >> if everything we need is exposed to the world or if we >> have to modify the runtime itself (which I would avoid if possible). > > I agree with this goal in principle, but: > > (1) Depending on what you want to do, it may be quite difficult to > do this, and > > (2) The RTS is actually quite hackable (as long as you are not > changing any of the interfaces with generated Haskell code, you > could even build your own RTS and have users link against that.) > > Something to keep in mind. > Edward, thanks for the quick reply! Yes, I see that the source code is clear and well documented! :) However, this project is something we would eventually want to publish on hackage, not something of internal use. This means we'd have to contribute the runtime changes upstream, which would be cool but I assumed GHC as a project doesn't accept every experimental new feature that come across, and we don't have the energy to guarantee maintenance of this eventual contribution. An external solution would be better. > > Not easily. If you make a safe foreign call, the capability is given up > before the C code executes, so you actually lose ownership of the > capability and TSO by the time you're running C code. The only > way to write a foreign call to properly suspend the thread that called > it is a primop function (e.g. stg_putMVar), and the operation is quite > delicate and you will probably want to use some existing (internal) > code in the RTS to do it. > > But if you just need to sleep/awake threads, why not just use an MVar? We don't have measurements, but we ruled out this possibility for performance reasons. Our idea is to make a thin Haskell wrapper around a tiny bit of highly optimized C code. What's the performance of locking on MVars? While we are at it: are primops callable directly from C? I suppose calling conventions are different. A question comes to mind: you mentioned "safe" calls. Are unsafe calls different regarding the detaching of the capability? Also: would a patch to make this possible be accepted? In particular: - add a way to make a "ultraunsafe" foreign call that do not loose the ownership of the calling thread. - add a BlockedOnExplicitSleep flag for tso->why_blocked (And in turn this means managing a different blocking queue, right?) - export something to reliably put in sleep and resume threads in this way. Is this feasible? Would it be a good idea? > > Edward Thank you again, Greetings, Nicola

3 4