[ghc-steering-committee] Status

Iavor Diatchki iavor.diatchki at gmail.com
Tue Mar 9 20:34:13 UTC 2021 

On Tue, Mar 9, 2021 at 12:22 PM Richard Eisenberg <rae at richarde.dev> wrote:

>
>
> On Mar 9, 2021, at 3:12 PM, Iavor Diatchki <iavor.diatchki at gmail.com>
> wrote:
>
> I still don't see what `sizeOf Bool` has to do with dependent
> types---there does not appear to be a semantic difference between
> `sizeOf @Bool` and `sizeOf Bool`, rather it is just a notational
> difference.
>
>
> Yes -- it's exactly the notational difference that's at issue. It's all a
> question of whether we want to require users to write @ before a required
> dependent (or type-level) argument. That really is it, but a good choice on
> this narrow (for me) hinges on whether or not we are working toward a
> future with dependent types.
>

For what it's worth I find `sizeOf @Bool` considerably more readable than
`sizeOf Bool`.




>  I am really missing the vision of where folks are picturing us ending up
> with all these changes.   Is the thinking that anyone who wants to use the
> FFI should have to enable dependent types?
>
>
> No. It's possible they would need no extensions at all. If there is no
> data constructor in scope named Bool, then sizeOf Bool might be accepted
> without any extensions on, I think. We often require an extension only at
> the definition site, not at usage sites.
>
> Does this help?
>
>
Well, it sounds like you are saying that to use FFI one would have to buy
into DH design.  My concern is not about what pragmas the users have to
write---after all we could make them write no pragams whatsoever.  It is
about what users would have to understand. For example, why does it matter
that there is no data constructor `Bool` in scope?   I am passing a type to
`sizeOf`, why would GHC all of a sudden get confused between the type and
the value?




> On Tue, Mar 9, 2021 at 11:33 AM Richard Eisenberg <rae at richarde.dev>
> wrote:
>
>> I agree with Arnaud. The connection to dependent types is pretty
>> tangential, really: it's all about whether we have `sizeOf @Bool` or
>> `sizeOf Bool`. If we had no intention of dependent types, I'd probably
>> favor the former. But forward-compatibility with the design sketch for
>> dependent types (
>> https://gitlab.haskell.org/ghc/ghc/-/wikis/dependent-haskell) suggests
>> using the latter. The problem is that the latter choice is more challenging
>> to support, and indeed drives much of the complexity in the proposal.
>>
>> So, we are a bit stuck: do we favor the design that's easier to specify
>> and implement today? or do we favor the design that will mesh better
>> tomorrow (assuming tomorrow has dependent types)? Our answer to #378 will
>> inform this decision.
>>
>> Richard
>>
>> On Mar 9, 2021, at 3:02 AM, Spiwack, Arnaud <arnaud.spiwack at tweag.io>
>> wrote:
>>
>> Simon's suggestion sounds like the proper thing to do to me too.
>>
>> The connection between #281 and dependent types is a little superficial,
>> but it #281 mixes the type syntax and the term syntax a bit more (in fact,
>> most of the details in the proposal are about that). The ability to mix
>> types and terms some way or another is central to having more dependent
>> typing. So there is, at the very least, a need for synchronization between
>> the various corresponding proposals. Though, in fact, this visible forall
>> thing is part of the dependent type package that Richard has been working
>> towards. It does fit somewhat naturally in the big picture.
>>
>> On Tue, Mar 9, 2021 at 6:58 AM Iavor Diatchki <iavor.diatchki at gmail.com>
>> wrote:
>>
>>> I will wait a few more days to see if anyone else expresses an opinion,
>>> otherwise I'll do as Simon suggests.
>>>
>>> Could someone clarify what does #281 have to do with dependent types?
>>>  The example that was brought up, and I think is pretty reasonable, is
>>> `sizeOf` from the `Storbale` class:  I might want to write `sizeOf @Bool`.
>>> Is the plan to require that dependent types are enabled to write stuff like
>>> that?
>>>
>>> -Iavor
>>>
>>>
>>>
>>> On Mon, Mar 8, 2021 at 8:35 PM Richard Eisenberg <rae at richarde.dev>
>>> wrote:
>>>
>>>> I have now submitted #378 for a committee decision (though I will
>>>> forgive Joachim if he missed the notification among the 200 comments on
>>>> that thread). So perhaps we can tackle all of this together.
>>>>
>>>> Richard
>>>>
>>>> On Mar 8, 2021, at 8:54 AM, Simon Peyton Jones via
>>>> ghc-steering-committee <ghc-steering-committee at haskell.org> wrote:
>>>>
>>>> I don't think this proposal is the right place to discuss dependent
>>>> types, so I propose that we reject the proposal.  To ensure progress, I'll
>>>> do so within a week, so please discuss before then, if you think otherwise.
>>>>
>>>> I would suggest that rather than “reject” (which sounds like “we don’t
>>>> want this, don’t resubmit it”), we push it back saying that we agree that
>>>> it’d be better to settle #378 first.  So we we’ll mark it as “needs
>>>> revision” not because we have anything specific to suggest, but so that
>>>> it’s back in the author’s control.  He should feel free to re-submit
>>>> (perhaps revised in some way, as he sees fit) once #378 is resolved.
>>>>
>>>> Simon
>>>>
>>>> *From:* Iavor Diatchki <iavor.diatchki at gmail.com>
>>>> *Sent:* 06 March 2021 16:51
>>>> *To:* Simon Peyton Jones <simonpj at microsoft.com>
>>>> *Cc:* Joachim Breitner <mail at joachim-breitner.de>;
>>>> ghc-steering-committee <ghc-steering-committee at haskell.org>
>>>> *Subject:* Re: [ghc-steering-committee] Status
>>>>
>>>>
>>>> Hello,
>>>>
>>>>
>>>>
>>>> This is the author's response to my query:
>>>>
>>>>
>>>>
>>>> So, the situation seems to be as follows:
>>>>
>>>>    - if we want dependent types (as described on that Wiki page), I
>>>>    would like to resubmit
>>>>
>>>>
>>>>    - if we do not want them, I will retract the proposal and close it
>>>>
>>>> I need to know the committee’s stance on dependent types to proceed
>>>> here.
>>>>
>>>>
>>>> I don't think this proposal is the right place to discuss dependent
>>>> types, so I propose that we reject the proposal.  To ensure progress, I'll
>>>> do so within a week, so please discuss before then, if you think otherwise.
>>>>
>>>> -Iavor
>>>>
>>>>
>>>>
>>>> On Fri, Mar 5, 2021 at 5:02 PM Iavor Diatchki <iavor.diatchki at gmail.com>
>>>> wrote:
>>>>
>>>> OK, I posted a note on the git-hub, asking @int-index what he wants to
>>>> do.
>>>> In case anyone needs a refresher, this is the discussion we had when I
>>>> submitted this to the committee back in November:
>>>>
>>>>
>>>> https://mail.haskell.org/pipermail/ghc-steering-committee/2020-November/001859.html
>>>> <https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmail.haskell.org%2Fpipermail%2Fghc-steering-committee%2F2020-November%2F001859.html&data=04%7C01%7Csimonpj%40microsoft.com%7C0b9f9dd78f934d620c5e08d8e0c0012f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637506462581487239%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=2inDHN90f7OQ9XIcd76l50Lwwp6axEUZVdea%2BoRFZzg%3D&reserved=0>
>>>>
>>>> another somewhat relevant discussion that followed was this:
>>>>
>>>> https://mail.haskell.org/pipermail/ghc-steering-committee/2020-November/001865.html
>>>> <https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmail.haskell.org%2Fpipermail%2Fghc-steering-committee%2F2020-November%2F001865.html&data=04%7C01%7Csimonpj%40microsoft.com%7C0b9f9dd78f934d620c5e08d8e0c0012f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637506462581497234%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=k4jiWTo11nNFH6zHyw7oYMFfyl0kxkQEsXKSueeC2Mk%3D&reserved=0>
>>>>
>>>> and here is a link to the proposal and the git-hub discussion:
>>>> https://github.com/ghc-proposals/ghc-proposals/pull/281
>>>> <https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fghc-proposals%2Fghc-proposals%2Fpull%2F281&data=04%7C01%7Csimonpj%40microsoft.com%7C0b9f9dd78f934d620c5e08d8e0c0012f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637506462581497234%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=X7tkk61XrEd9utRxcQjA3Eh3Wi3cDXuJdMae%2BBcx3gM%3D&reserved=0>
>>>>
>>>> -Iavor
>>>>
>>>>
>>>>
>>>> On Fri, Mar 5, 2021 at 3:13 PM Simon Peyton Jones <
>>>> simonpj at microsoft.com> wrote:
>>>>
>>>> This is about #281 visible dependent quantification.
>>>>
>>>> Our role is to accept, reject, or push back to the author for
>>>> revision.  We should do so in a timely way.
>>>>
>>>> If I were the shepherd I’d be asking the author “Do you want to revise
>>>> the proposal at all in the light of all the discussion, or do you want the
>>>> committee to decide yay/nay based on the proposal as it is now?”
>>>>
>>>> He might want to revise/discuss a bit.  But if he says “please decide
>>>> on it as-is” then I think you should make a recommendation: accept or
>>>> reject.  (I don’t think the discussion has led to any specific revisions
>>>> that we want to see.)
>>>>
>>>> I sense you would recommend reject.   That’s totally fine. Then the
>>>> rest of us have to make up our minds.  We might end up with a more nuanced
>>>> position like “we are not ready to accept this now, so if you want a
>>>> decision now, it’d be no; but feel free to ask the same question again in
>>>> six months time”.
>>>>
>>>> If you don’t feel able to be the shepherd, that’s fine too: tell
>>>> Joachim and he’ll finger someone else!
>>>>
>>>> But, by hook or by crook, I do think we should do something, not just
>>>> sit on it.  Tom’s nudge is helpful.
>>>>
>>>> Simon
>>>>
>>>> *From:* Iavor Diatchki <iavor.diatchki at gmail.com>
>>>> *Sent:* 05 March 2021 19:42
>>>> *To:* Simon Peyton Jones <simonpj at microsoft.com>
>>>> *Cc:* Joachim Breitner <mail at joachim-breitner.de>;
>>>> ghc-steering-committee <ghc-steering-committee at haskell.org>
>>>> *Subject:* Re: [ghc-steering-committee] Status
>>>>
>>>>
>>>> I am being nudged to do something with this proposal,  but I am not
>>>> sure what...   Is the rest of the committee OK with moving it back
>>>> "revisions required", and if so what revisions would we like?
>>>>
>>>>
>>>>
>>>> I would be quite happy if someone else wanted to shephard this.  As I
>>>> mentioned before, I don't have much interest in DH, so I have not followed
>>>> the really long discussion related to that, and if and how it might relate
>>>> to #281.
>>>>
>>>>
>>>>
>>>> My recommendation for the moment would be:
>>>>
>>>>    * This seems like a useful feature independent of DH, so it would be
>>>> nice to come up with a concise notation to use the feature on its own,
>>>> without worrying about DH.
>>>>
>>>>
>>>>
>>>> Please let me know what you think, as I am not sure what is the
>>>> committee's stance on the proposal.
>>>>
>>>>
>>>>
>>>> -Iavor
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Thu, Feb 18, 2021 at 1:54 PM Simon Peyton Jones <
>>>> simonpj at microsoft.com> wrote:
>>>>
>>>> There has been a broad further discussion.
>>>> What we advertise is that, rather that leave a proposal “under
>>>> committee review”, we will push it back to the author with an invitation to
>>>> resubmit when the discussion has died down and they feel ready to submit a
>>>> proposal, revised in the light of the discussion.  That’s different to
>>>> reject… it means that there is an ongoing debate so it’s not a good time
>>>> for the committee to make a decision.
>>>>
>>>> Simon
>>>>
>>>> *From:* ghc-steering-committee <
>>>> ghc-steering-committee-bounces at haskell.org> *On Behalf Of *Iavor
>>>> Diatchki
>>>> *Sent:* 18 February 2021 16:47
>>>> *To:* Joachim Breitner <mail at joachim-breitner.de>
>>>> *Cc:* ghc-steering-committee <ghc-steering-committee at haskell.org>
>>>> *Subject:* Re: [ghc-steering-committee] Status
>>>>
>>>>
>>>> On #381 I think the idea of visible quantification makes sense every
>>>> now and then, but I don't like the concrete details of the proposal: the
>>>> magic lifting of terms to types seems quite complicated, and using `type`
>>>> as an explicit herald doesn't look nice.  So I don't think it's the right
>>>> design, and therefore I suggest we reject the proposal.
>>>>
>>>>
>>>>
>>>> I am sure that others would disagree as apparently this is an essential
>>>> part of dependent Haskell.   I have not followed the large discussion that
>>>> Richard created, as I am not particularly interested in the design being
>>>> proposed, so perhaps someone else should champion this.
>>>>
>>>>
>>>>
>>>> Aslo, I am not sure if I am actually on the committee, as I thought my
>>>> term had expired?  That might be more reason for someone else to pick it up.
>>>>
>>>>
>>>>
>>>> -Iavor
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Thu, Feb 18, 2021 at 2:32 AM Joachim Breitner <
>>>> mail at joachim-breitner.de> wrote:
>>>>
>>>> Dear Committee,
>>>>
>>>> another status update, because why not.
>>>>
>>>> A small reminder: These mails have two sections, one that’s a delta
>>>> since last status, but below is a summary of all proposals we have to
>>>> act on. Please at least scroll through that on each status mail to see
>>>> if you are listed, maybe you forgot something (or I made a mistake).
>>>>
>>>> So here’s the delta since last month.
>>>>
>>>>  * GHC2021 defined! Yay!
>>>>
>>>>  * Bylaws merged! Yay
>>>>
>>>>  * Simon², Iavor, Richard and me had thus their terms expired.
>>>>
>>>>    The Simons, as key members, wanted to continue and were
>>>>    voted back in.
>>>>
>>>>    The others are now “expiring”, until the next nomination round
>>>>    concludes. Alejandro is going to run that process.
>>>>
>>>>  * we were asked to review these proposals:
>>>>    #390: Fine-grained pragmas, Shepherd: Vitaly
>>>>
>>>>  * we have a recommendation from the shepherd about:
>>>>    #368: Warn on prefix/suffix operators (accept)
>>>>
>>>>  * we have sent the following proposals back to revision
>>>>    - none -
>>>>
>>>>  * we decided about the following proposals
>>>>    #313: Delimited continuation primops (accept)
>>>>    #387: The Char kind (accept)
>>>>    #368: Warn on prefix/suffix operators (accept)
>>>>
>>>> We currently have to act on the following 5 proposals, down by 2.
>>>>
>>>> ## Waiting for committee decision
>>>>
>>>> #381: Visible 'forall' in types of terms, Shepherd: Iavor
>>>>       Recommendation was to reject, but discussion went into the more
>>>>       abstract “whither dependent Haskell”. But what does this mean
>>>>       for this proposal?
>>>>       Iavor, can you pick this up again?
>>>>
>>>> #369: Add sumToTag# primop, Shepherd: Eric
>>>>       Essentially accepted, waiting for feedback from the author on
>>>>       final tweaks. Eric, care to nudge the author, or just do it?
>>>>
>>>> #302: \of, Shepherd: Cale
>>>>       No new discussion yet. It seems there was some confusion, which
>>>>       was cleared up by Tom, and Cale said he’ll pick it up now again.
>>>>
>>>>
>>>> ## Waiting for Shepherd action
>>>>
>>>> #367: Clarify primops using unboxed sums, Shepherd: Simon Marlow
>>>>       Simon said he’d reject it on the Github PR. Still waiting
>>>>       for the discussion to start on the mailing list.
>>>>
>>>> #390: Fine-grained pragmas, Shepherd: Vitaly
>>>>       Still kinda new, but a recommendation would be good soon.
>>>>
>>>>
>>>> Cheers,
>>>> Joachim
>>>> --
>>>> Joachim Breitner
>>>>   mail at joachim-breitner.de
>>>>   http://www.joachim-breitner.de/
>>>> <https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.joachim-breitner.de%2F&data=04%7C01%7Csimonpj%40microsoft.com%7C0b9f9dd78f934d620c5e08d8e0c0012f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637506462581507230%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=gsdKkpp3AhJz2ins9uuctT%2FGf6rr2bScSHFWyX3jFhM%3D&reserved=0>
>>>>
>>>>
>>>> _______________________________________________
>>>> ghc-steering-committee mailing list
>>>> ghc-steering-committee at haskell.org
>>>> https://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-steering-committee
>>>> <https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmail.haskell.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Fghc-steering-committee&data=04%7C01%7Csimonpj%40microsoft.com%7C0b9f9dd78f934d620c5e08d8e0c0012f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637506462581517226%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=BY04FgtL9RHpjI6frr%2BOYfxwPkW5yjN2DeRyv2JaKN4%3D&reserved=0>
>>>>
>>>> _______________________________________________
>>>> ghc-steering-committee mailing list
>>>> 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
>>> 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/20210309/a97c71f5/attachment-0001.html>

More information about the ghc-steering-committee mailing list