August 2018 - ghc-steering-committee

Bundling patterns with type synonyms (#28)
by Christopher Allen 26 Jan '19

26 Jan '19

The proposal I'm reviewing is: Bundling patterns with type synonyms by Bertram Felgenhauer and Joe Hermaszewski https://github.com/ghc-proposals/ghc-proposals/pull/28 Having reviewed the proposal and gone through some of the GHC Trac history, I think we should accept this proposal. There are a couple of open questions, ambiguities, and potential downsides but the overall complexity of the proposal doesn't cause me great concern. Cheers, Chris Allen

5 11

Proposal: Top-level kind signatures (#54)
by Roman Leshchinskiy 30 Sep '18

30 Sep '18

Hi everyone, The proposal is about adding stand-alone kind signatures for type declarations: https://github.com/ghc-proposals/ghc-proposals/pull/54 The basic idea is to add support for kind signatures like this: type T :: Type -> Type -> Type data T a b = ... Signatures would be introduced by the keyword 'type' and could be provided for data types, type synonyms, type families and classes. The extension would be enable via -XTopLevelKinds. Additionally, complete user-supplied kind signatures (CUSKs, see https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_ext…) would be given a separate language extension, -XCUSKs which would be on by default for now but turned off in GHC 8.8 and then deprecated and removed. The proposal depends on #81. Feedback has been positive and this is rather cleaner than the current story so I'm in favour of accepting this. Like Richard, I'm not really happy with the name "TopLevelKinds", especially since signatures for associated types aren't even on the top level. Would something like StandAloneKinds or ExplicitKindSignatures be better? One minor point that isn't clear to me: the proposal says that "Associated types may be given kind signatures within their classes". Are type variables mentioned in the class head visible in the kind signature for the associated type? Thanks, Roman

4 3

Proposal: Syntax for visible dependent quantification (#81)
by Roman Leshchinskiy 30 Sep '18

30 Sep '18

Hi everyone, The proposal is about adding support for dependent quantification to kind signatures: https://github.com/ghc-proposals/ghc-proposals/pull/81 Consider the following declaration (example lifted from the proposal): data T k (a :: k) GHC accepts this but it can't be given an explicit kind. Internally, it is assigned a kind which is rendered as forall k -> k -> * but this isn't accepted in source code. Note that in applications of T, k must be specified explicitly (e.g., T Type Int) which is why T does *not* have the kind forall k. k -> * Moreover, k is mentioned later in the kind which is why something like Type -> k -> * doesn't work, either. The proposal is to allow forall k -> k -> * and similar kinds to appear in source code. This is actually intended as the first in a series of proposals driving us towards dependent types in Haskell as described in Richard's thesis (https://www.cis.upenn.edu/~sweirich/papers/eisenberg-thesis.pdf) Ultimately, the intention is to have all of the following (cf. Chapter 4 of the thesis): forall a. t forall a -> t pi a. t pi a -> t Here, forall and pi express relevance (does it exist at runtime) and . and -> express visibility (does it have to be specified explicitly). Because of this, my recommendation is to strongly encourage the author to submit an extended proposal which reserves (but doesn't specify the semantics of) the above syntax wholesale. This would allow us to ensure that various bits of Dependent Haskell use consistent syntax and language extensions once implemented. I find it quite difficult to discuss just this specific bit of syntax in isolation. Indeed, the public discussion was rather confused without an explanation of the roadmap (https://github.com/ghc-proposals/ghc-proposals/pull/81#issuecomment-3368929…) Alternatively, we could just agree on the roadmap ourselves, without public discussion. This would somewhat circumvent the process, though. If we decide to discuss just the proposal as is, though, then I'd be weakly against the proposed syntax as it is too subtle for my taste and abuses familiar mathematical notation somewhat. I'd probably prefer something like: type a -> t The proposal also doesn't specify what language extension would turn on support for the syntax so this would have to be rectified. Thanks, Roman

4 9

Proposal: Lazy unboxed tuples / warn on unbanged strict patterns (#35); Recommendation, attempt 2: Accept
by Vitaly Bragilevsky 12 Sep '18

12 Sep '18

Hi, This is my revised recommendation for the lazy unboxed tuples proposal ( https://github.com/ghc-proposals/ghc-proposals/pull/35) It is already implemented, and I don't see strong opposition to it, so my recommendation is ACCEPT. Silence is understood as agreement. Regards, Vitaly

2 2

Proposal: Define Kinds Without Promotion (#106)
by Eric Seidel 12 Sep '18

12 Sep '18

Hi everyone! This proposal would allow the declaration of new kinds and type constructors, without the associated data constructors that would have been generated by -XDataKinds. https://github.com/yav/ghc-proposals/blob/data-kind-only/proposals/0000-typ… It introduces a new language extension, -XTypeData, which allows declarations like the following: type data Universe = Character | Number | Boolean This introduces a new kind, Universe, with three associated type constructors Character :: Universe Number :: Universe Boolean :: Universe Notably, no data constructors are introduced. The proposal aims to solve several usability issues with -XDataKinds: 1. We often don't want the data constructors, which simply pollute the value namespace. 2. We cannot easily control the export behavior of promoted types (instead we resort to tricks like the alias `type Character = 'Character`). 3. The name resolution rules involving promoted types can be confusing. I find these issues (particularly #2) compelling, and the community response seems to be mostly positive, so I recommend accepting the proposal. I'm not particularly fond of the proposed syntax (`type data` makes it sound like we're doing something with data constructors, which is precisely what we're NOT doing), but the proposal has already had a good bit of bikeshedding, and people seem mostly happy with the current syntax, so I think we should probably stick with `type data`. Eric

6 24

Please Review: StableName guarantee (#163), Shepherd: Simon Marlow
by Joachim Breitner 30 Aug '18

30 Aug '18

Dear Committee, this is your secretary speaking: Remove an undocumented StableName guarantee, by David Feuer, was proposed https://github.com/ghc-proposals/ghc-proposals/pull/163 I propose Simon Marlow as the shepherd. Please reach consensus as described in https://github.com/ghc-proposals/ghc-proposals#committee-process I suggest you make a recommendation, in a new e-mail thread with the proposal number in the subject, about the decision, maybe point out debatable points, and assume that anyone who stays quiet agrees with you. Thanks, Joachim -- Joachim Breitner mail(a)joachim-breitner.de http://www.joachim-breitner.de/

1 0

Proposal #111: Linear Types
by Richard Eisenberg 29 Aug '18

29 Aug '18

I have volunteered to shepherd Proposal #111: Linear Types. The pull request is here: https://github.com/ghc-proposals/ghc-proposals/pull/111 The authors propose adding linear functions to GHC. The argument to a linear function must be used exactly once in the function body. Motivation for this feature abounds -- essentially, linear types allow for better static checking around resource allocation and deallocation. I refer you to the proposal (and published paper) for more details here. The proposal also contains multiplicity polymorphism, where a higher-order function can be polymorphic in the multiplicity of a supplied function. Datatypes are linear by default, with a long list of (in my opinion, non-obvious) rules around syntax and exceptions to those rules. Constructors have a different multiplicity when used in an expression than they do when used in a pattern, for example. The authors leave type inference as an open question and do not state how the feature would work with either view patterns or pattern synonyms. The proposal comes with a companion document detailing the changes to Core. These changes seem appropriate. The current proposal has a fair number of open questions. I've written a long comment on the GitHub trail detailing what I was uncertain of. The proposal is not currently near the level of crispness of the proposals we normally accept. For a smaller proposal, I would just send it back to the authors without really giving it to the committee. However, there has been a great deal of effort behind writing this proposal and building an implementation, and so I think it's only fair that we give some thought about the general direction being worked in here. As to the features themselves: I'm personally far from convinced. Though I see no better way to achieve the authors' goals, the feature they have proposed is full of both sharp and dark corners. If implemented, there would be regular questions cropping up as to why such-and-such is the case or why the design is the way it is. However, this is well motivated and eagerly anticipated. And I don't think we'll be able to conjure up a better design without gaining experience with the proposed design, however flawed. I thus recommend: Encouragement to continue the proposal and implementation, with an eye toward acceptance. Furthermore, if/when merged, I would like to advertise that the feature is subject to change, with no backward compatibility guarantee, for several versions. If the feature is implemented and merged, we will learn more about it and then perhaps refine it in the future. Richard

8 25

Discussion about "Type Application in Patterns" (#126)
by Iavor Diatchki 20 Aug '18

20 Aug '18

Hello, let's also start the discussion on feature request 126. The idea here is that we allow the @ notation for explicit type applications to also be used on constructors in patterns. Using @ with a constructor in a pattern has the same meaning as it does it an expression: the provided type is used to instantiate the corresponding type parameter of the constructor. If the type contains variables, those are treated in the same way as in #128, where "unbound" variables name the matching types. Here are some examples: f1 (Just @Int x) = x -- This has type `Maybe Int -> Int` f2 (Just @[a] x) = x == "c" -- `a` is an alias for `Char` f3 (SomeException @e ex) = ... -- `e` is a name for the existentially hidden exception type Overall I think that is a simple and natural extension to the way @ already works, and I propose that we accept it. Thoughts? -Iavor

6 23

Proposal: Lazy unboxed tuples / warn on unbanged strict patterns (#35); Recommendation: Reject
by Vitaly Bragilevsky 20 Aug '18

20 Aug '18

Hi, the lazy unboxed tuples proposal ( https://github.com/ghc-proposals/ghc-proposals/pull/35) was under discussion for a long period of time (more than a year and a half since submission). As a shepherd to this proposal I recommend rejection based on the following: * there is no clearly articulated motivation in favor of this proposal despite complying with the Manual; * implementing this would lead to hard-to-trace performance issues in the users code (due to move from strictness in current GHC behaviour to laziness); * it looks like the change (B) of the proposal (warn an unbanged strict patterns) meets no complains, so it is better to be resubmitted as a separate proposal; * if resubmitted separately the change (A) should elaborate on desugaring to make potential performance drawbacks clear; * it seems (maybe mistakenly) that the author has lost his interest in this proposal. Although silence is undestood as agreement, I'd be glad to receive a feedback on this recommendation. Regards, Vitaly

5 7

Status
by Joachim Breitner 17 Aug '18

17 Aug '18

Hi committee, with the new nominations out of the way, time to stop navel gazing and doing some work! Since the last status update, we * Got two new members. Yay and welcome! * were asked to review these proposals: #149 Replace atomicModifyMutVar# (Shepherd: Simon M) #143 Remove * syntax (Shepherd: Ben) #128 Allow ScopedTypeVariables refer to types (Shepherd: Iavor) #126 Type Applications in Patterns (Shepherd: Iavor) * two reviews were reassigned: #106 Define Kinds Without Promotion (New shepherd: Eric) #35 Lazy unboxed tuples (New shepherd: Vitaly) * got a recommendation from shepherds about: #112 Remove TypeOperators => NoStarIsType "migration" (rec: accept) #149 Replace atomicModifyMutVar# (rec: accept) #111 Linear Types (rec: acceptish) #143 Remove * syntax (rec: accept) #128 Allow ScopedTypeVariables refer to types (rec: accept) #126 Type Applications in Patterns (rec: accept) * decided about the following proposals #99 forall {k} (accept) #112 Remove TypeOperators => NoStarIsType "migration" (accept) #149 Replace atomicModifyMutVar# (accept) #134 Deprecating Exports (accept) #128 Allow ScopedTypeVariables refer to types (accept) #143 Remove * syntax (accept) We currently have to act on the following 11 proposals, which is two less than last round. Lets further get the backlog down! Type applications in patterns https://github.com/ghc-proposals/ghc-proposals/pull/126 Shepherd: Iavor Status: Overall positive response, acceptance likely Type annotated quoters https://github.com/ghc-proposals/ghc-proposals/pull/125 Shepherd: Manuel Status: Still waiting for recommendation. Provenance-Qualified Package Imports https://github.com/ghc-proposals/ghc-proposals/pull/115 Shepherd: Ben Status: Still waiting for recommendation. This is pretty old! Linear types https://github.com/ghc-proposals/ghc-proposals/pull/111 Shepherd: Richard Status: acceptance-ish(?) Define Kinds Without Promotion https://github.com/ghc-proposals/ghc-proposals/pull/106 Shepherd: Eric Status: Reassigned to Eric, waiting for recommendation. ExtraCommas https://github.com/ghc-proposals/ghc-proposals/pull/87 Shepherd: Chris Recommendation: accept Status: Ongoing discussion, met with some reservation. Visible dependent quantification (TL;DR: forall k -> k -> *) https://github.com/ghc-proposals/ghc-proposals/pull/81 Shepherd: Roman Leshchinskiy Status: Discussion needs to happen Top-level kinds signatures https://github.com/ghc-proposals/ghc-proposals/pull/54 Shepherd: Roman Leshchinskiy Status: Waiting for recommendation. Or-Patterns https://github.com/ghc-proposals/ghc-proposals/pull/43 Shepherd: Manuel Recommendation: accept Status: Round 2 discussion ebbed down Lazy unboxed tuples https://github.com/ghc-proposals/ghc-proposals/pull/35 Shepherd: Vitaly Status: Reassigned to Vitaly, waiting for recommendation. Bundling patterns with synonyms https://github.com/ghc-proposals/ghc-proposals/pull/28 Shepherd: Chris Recommendation: accept Status: Ongoing discussion, met with some resistance. Cheers, Joachim -- Joachim Breitner mail(a)joachim-breitner.de http://www.joachim-breitner.de/

2 1