[ghc-steering-committee] Unsaturated type families (#242)

Wed Dec 2 16:41:22 UTC 2020

I felt rather too biased to comment on this proposal, as I have been very publicly vocal about how excited I am to see it in GHC. However, it occurs to me that my thoughts on the proposal are rather too scattered across the Internet, so I’ll try to collate the main points that I have. I will caveat these by saying that they have mostly been addressed in the thread and/or the proposal, and they certainly don’t change my opinion that the proposal be accepted.

* A pretty significant consequence of this proposal is partial application of type synonyms. I think the inability to partially apply these is a rather counter-intuitive restriction to beginners (particularly when they’ve just started to get familiar with partial application at the value level). Certainly, when introducing colleagues to Haskell, it is a question that comes up reasonably often: “why must I write type T = Maybe and never type T x = Maybe x?” Of course, once one considers the importance of `f a ~ g b <=> (f ~ g, a ~ b)` for inference, and an example like type T x = Either x Int, this starts to make more sense, but we’ve had to reach a surprisingly deep level of understanding to figure this out.

Naturally, lifting that restriction is going to mean one less thing to explain ("type-level works like term-level!”) but I worry about the potential error messages here. If a user is at a point where they feel comfortable playing with type families, I think they’re perhaps more primed for strange error messages, but can I ever run into a matchability mismatch error as an absolute beginner? The classic example (that Joachim already raised) is trying to write a functor instance for type Id x = x. If I’m new to the language, it’s potentially not clear to me why I can’t do this (at least until the rest of Csongor’s research finds its way into the compiler). In which case, I’d like to see a more specialised error such as, “You can’t write a class instance for a partially-applied type synonym”, rather than exposing someone to 'M and 'U in their first week of Haskell.

* Now we have unmatchable (not generative, not injective) and matchable (generative, injective), it feels incomplete not to have annotations for generative-but-not-injective and injective-but-not-generative arrows. While generativity annotation (and checking!) is effectively an entirely new chunk of work, it’s a shame that we have the syntax to specify the injectivity of a type family when we create it, but not when we consume it. I have some concrete examples of where having to declare injective type families as fully unmatchable leads to ambiguity that I can share if anyone is interested, but my understanding from Csongor is that this is a deceptively large amount of work, and definitely out of scope for this particular proposal.

Looking down my list, these are the two thoughts I have that haven’t been explicitly covered in full somewhere in the GitHub thread, and I feel they could be summarised as an implementation suggestion and a feature request. All of which is to say, I’m still very much in support of the proposal, and extremely excited to start using it.

Thanks,
Tom

On 2 Dec 2020, at 15:30, Spiwack, Arnaud <arnaud.spiwack at tweag.io<mailto:arnaud.spiwack at tweag.io>> wrote:

Dear all,

I'm a bit worried by the limited response from the committee on this proposal. It is a non-trivial proposal, and I think it deserves more eyes. So please have a look at them so that we can commit to this with confidence.

/Arnaud

On Tue, Dec 1, 2020 at 7:28 PM Richard Eisenberg <rae at richarde.dev<mailto:rae at richarde.dev>> wrote:
This thread has seen only positive responses, and with no responses for the past 6 days. I'm thus inclined to accept the proposal.

However, there are open questions around the following points:
* the concrete syntax (pending discussion on #370)
* defaulting rules (as raised by Alejandro in this thread)

Conveniently, these are both listed as Unresolved Questions in the proposal itself. We need a way of resolving these questions. The syntax question may become clearer once we know what to do about #370. The defaulting question is harder. I vote to return to this question once #378 has settled somewhat -- but even then, it will be hard. Still, I think we should move forward with accepting the main proposal, and we can continue to debate the defaulting strategy in a further thread, perhaps in parallel with reviewing the (already existing) implementation.

I will accept this proposal as written at the end of the week, barring commentary here (or on GitHub) to stop me.

Thanks!
Richard

> On Nov 25, 2020, at 10:28 AM, Simon Peyton Jones <simonpj at microsoft.com<mailto:simonpj at microsoft.com>> wrote:
>
> I'm on board with unsaturated type families.
>
> Like Linear Haskell, it's quite a "big" proposal, but accepting it is compatible with idea of Haskell as a laboratory for exploration.  I think we should flag it as experimental, with the implication that details are liable to change as we gain experience.
>
> Like others, I'd like us to converge on #370 before fixing syntax.
>
> I don’t have a strong opinion about the defaulting stuff.
>
> Declaration of interest: I'm a co-author on the paper.
>
> Simon
>
> |  -----Original Message-----
> |  From: ghc-steering-committee <ghc-steering-committee-
> |  bounces at haskell.org<mailto:bounces at haskell.org>> On Behalf Of Richard Eisenberg
> |  Sent: 20 November 2020 19:36
> |  To: Simon Peyton Jones via ghc-steering-committee <ghc-steering-
> |  committee at haskell.org<mailto:committee at haskell.org>>
> |  Subject: [ghc-steering-committee] Unsaturated type families (#242)
> |
> |  Hi committee,
> |
> |  Csongor Kiss has proposed -XUnsaturatedTypeFamilies.
> |
> |  Proposal:
> |  https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
> |  ub.com<http://ub.com/>%2Fkcsongor%2Fghc-proposals%2Fblob%2Funsaturated-type-
> |  families%2Fproposals%2F0000-unsaturated-type-
> |  families.rst&data=04%7C01%7Csimonpj%40microsoft.com<http://40microsoft.com/>%7Cbd9e62e3137e
> |  40385b5f08d88d8b91e0%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C1%7C6374
> |  14978666235646%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2l
> |  uMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=WdLjCp2ReG9ZXOjmE
> |  Ow6VCJlAO7Yf1aWkVAXHxrsmMM%3D&reserved=0
> |  ICFP'19 paper:
> |  https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.
> |  microsoft.com<http://microsoft.com/>%2Fen-
> |  us%2Fresearch%2Fuploads%2Fprod%2F2019%2F03%2Funsaturated-type-
> |  families-icfp-
> |  2019.pdf&data=04%7C01%7Csimonpj%40microsoft.com<http://40microsoft.com/>%7Cbd9e62e3137e4038
> |  5b5f08d88d8b91e0%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C1%7C63741497
> |  8666235646%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzI
> |  iLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=dgowMlsNpH0X9fPAr498F
> |  Y9u8xML9n0G1nwvPN4R9HA%3D&reserved=0
> |  Discussion:
> |  https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
> |  ub.com<http://ub.com/>%2Fghc-proposals%2Fghc-
> |  proposals%2Fpull%2F242&data=04%7C01%7Csimonpj%40microsoft.com<http://40microsoft.com/>%7Cbd
> |  9e62e3137e40385b5f08d88d8b91e0%7C72f988bf86f141af91ab2d7cd011db47%7C1%
> |  7C1%7C637414978666235646%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiL
> |  CJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=C9N0y7l
> |  KhZsepv0mkXbJOMDVLFi4bN5kaVW7DbXW1ro%3D&reserved=0
> |
> |  The central idea is to allow type functions (both type families and
> |  type synonyms) to appear unsaturated. (Currently, all type functions
> |  are required syntactically to be applied to all parameters they are
> |  declared with.) This poses a problem for type inference, as detailed
> |  in both the proposal and the paper. The key question: if we have (f a
> |  ~ g b), can we conclude (f ~ g) and (a ~ b)? Not if either of f or g
> |  is a type function. This proposal thus describes a mechanism to
> |  introduce a new flavor of arrow, such that we can identify type
> |  functions by their kind. Specifically, we have regular types like
> |  Maybe :: Type -> @M Type (where the M stands for "matchable"), but
> |  type functions like F :: Type -> @U Type (where the U stands for
> |  "unmatchable"). Unmatchable applications can not be decomposed during
> |  type inference.
> |
> |  Much of the proposal is concerned with backward-compatibility: most
> |  users will not want to write @M or @U after each of their arrows, so
> |  the proposal describes ways of defaulting this behavior to match
> |  (most) programmers' expectations.
> |
> |  The proposal also includes matchability polymorphism, the ability to
> |  abstract over a matchability parameter.
> |
> |  Pros:
> |  + This proposal greatly increases the expressiveness of Haskell's type
> |  system.
> |  + With this proposal, we can finally do proper functional programming
> |  in types, rather than just in terms.
> |  + This proposal is a key ingredient toward having dependent types, as
> |  + unsaturated functions are commonplace in terms, and thus should also
> |  be supported in types. (Allowing unsaturated functions in types was a
> |  key difference between Adam Gundry's proposal for dependent types
> |  https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fadam
> |  .gundry.co.uk<http://gundry.co.uk/>%2Fpub%2Fthesis%2F&data=04%7C01%7Csimonpj%40microsoft
> |  .com%7Cbd9e62e3137e40385b5f08d88d8b91e0%7C72f988bf86f141af91ab2d7cd011
> |  db47%7C1%7C1%7C637414978666235646%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4w
> |  LjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdat
> |  a=wk7Xc%2Bb5FLSrndKYZ2ytJh6gO2oYiCXLDhQhdEOfSOg%3D&reserved=0,
> |  which requires a notion of a subset of the type and term languages
> |  shared in common, and mine, which makes no distinction between terms
> |  and types.) There is a prototype implementation.
> |  + The ideas are backed up by peer-reviewed research.
> |  + Despite focusing on type families, this work applies equally to
> |  ordinary functions which might be used in types once we have stronger
> |  support for dependent types.
> |
> |  Cons:
> |  - This adds a new dimension of complexity to our kind system, by
> |  separating out matchable and unmatchable arrows.
> |  - The rules for defaulting appear convenient in practice, but are
> |  somewhat arbitrary.
> |  - The rules for defaulting care about context -- does an arrow appear
> |  in the type of a term or the type of a type? These rules thus go
> |  against the spirit of #378, which advocates for not accepting features
> |  that distinguish between types and terms.
> |
> |  Recommendation: With reservations, I recommend acceptance. I think
> |  that the power to use higher-order programming should not be
> |  restricted to terms, and allowing unsaturated functions at compile
> |  time is necessary in order to have convenient dependent types.
> |  However, I am concerned about the extra complexity of matchability. A
> |  key open question for me is how much matchability is apparent to users
> |  -- even ones using some higher-order type-level programming. If
> |  matchability is pervasive, then I would lean against. But my
> |  expectation is that matchability fades into the background -- much
> |  like levity polymorphism (unless you want it).
> |
> |  Open question: What to do about syntax? The proposed syntax is
> |  sensible. However, #370 suggests an alternative syntax that might be
> |  more forward-thinking.
> |
> |  Richard
> |  _______________________________________________
> |  ghc-steering-committee mailing list
> |  ghc-steering-committee at haskell.org<mailto:ghc-steering-committee at haskell.org>
> |  https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmail
> |  .haskell.org<http://haskell.org/>%2Fcgi-bin%2Fmailman%2Flistinfo%2Fghc-steering-
> |  committee&data=04%7C01%7Csimonpj%40microsoft.com<http://40microsoft.com/>%7Cbd9e62e3137e403
> |  85b5f08d88d8b91e0%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C1%7C6374149
> |  78666235646%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMz
> |  IiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=cIlDCT4r8C8Yc0%2BOib
> |  Q%2F6Dv1qzBiB1PpavCdAJI3ruw%3D&reserved=0

_______________________________________________
ghc-steering-committee mailing list
ghc-steering-committee at haskell.org<mailto:ghc-steering-committee at haskell.org>
https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-steering-committee
_______________________________________________
ghc-steering-committee mailing list
ghc-steering-committee at haskell.org<mailto:ghc-steering-committee at haskell.org>
https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-steering-committee

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-steering-committee/attachments/20201202/656c609d/attachment-0001.html>