[ghc-steering-committee] Please review #220: QualifiedImports, Shepherd: Simon M

Tue Sep 17 22:56:22 UTC 2019

I agree with Arnaud that defaults are important: switching from opt-in to opt-out for organ donation, and for pension contributions, has had a huge effect.

Suppose, as in the above examples, we had a consensus that import-qualified-by-default was the way we wanted Haskell to be.  Then we’d just be discussing how to switch over, what the deprecation strategy is, how many compiler releases to allow etc.  Controversial as it was, the Foldable thing was like this.

But that’s not the case here.   We are not discussing a change to the base language, but a language extension that you can choose to use, or not.  So it remains a local coding style choice: “in our company we always use `{-# LANGUAGE QualifiedImports #-}` “.  But that really isn’t significantly different from saying “in our company we always use `import qualified`”.  Yes, you can put the former in your .cabal file – but you could equally use HLint to enforce qualified import.

I’m still not persuaded that this small change in convenience is enough to add an extension for.   As I say, it’d be different if there was a consensus that we wanted to change the base language, and migrate everyone to the new default.

Simon

From: ghc-steering-committee <ghc-steering-committee-bounces at haskell.org> On Behalf Of Richard Eisenberg
Sent: 17 September 2019 22:04
To: Spiwack, Arnaud <arnaud.spiwack at tweag.io>
Cc: ghc-steering-committee at haskell.org; Joachim Breitner <mail at joachim-breitner.de>; Sandy Maguire <sandy at sandymaguire.me>
Subject: Re: [ghc-steering-committee] Please review #220: QualifiedImports, Shepherd: Simon M

I think Arnaud's is an effective argument. It has not changed my opinion of this proposal (I'm still against), but it's moved the needle a bit for me. What it does suggest is a feature in HLint (if it doesn't already exist) that encourages this behavior. Putting this in HLint allows individuals and organizations to experiment with enforcing this style, perhaps building up further experience for retrying this in the future.

After writing the above, I talked to Arnaud, and he pointed out a deadly flaw in this plan: even if qualified is the default, it might sometimes be nice to import unqualified. In the proposal, this is done via the unqualified keyword. But without this proposal, there would be no way to signal to HLint that an import is meant to be unqualified. So perhaps a much more modest proposal allowing (but not requiring) users to write `unqualified` in import statements would be worthwhile. This keyword would always be redundant today, but it might feasibly be a way forward here.

Full disclosure: my immediate instinct would be to be against the "unqualified" proposal, but it wouldn't be a fork of the language (unlike the current proposal). Perhaps others could convince me otherwise.

Richard

On Sep 17, 2019, at 7:52 AM, Spiwack, Arnaud <arnaud.spiwack at tweag.io<mailto:arnaud.spiwack at tweag.io>> wrote:

Dear all,

As one of the author of this proposal. I am, unsurprisingly, against rejecting it. Though it seems I'm rather in a minority here, let me add one last argument to try and sway the general opinion. Being understood that being an author, this argument cannot, in any way be considered as “a vote” or any such thing.

Human psychology is powerful. As it happens, we have a very strong tendency to choose whatever course of thought or action requires the least mental effort. Defaults require very little mental efforts, so we naturally will gravitate towards default. This is why, for instance, almost every Swedish worker is part of a union, while almost every French worker isn't: in Sweden, unionising is opt-out, whereas in France, it's opt-in. That's also why putting apples in front of sweet deserts in a school restaurant will result in more children eating fruits rather than cakes.

Back to our case: the overwhelming majority of Haskell packages are designed to be used unqualified (and also do almost all of their imports unqualified). Now, either unqualified import are really that much better, or the default has an enormous influence. As I previously mentioned, in Ocaml, a fairly similar language, qualified is the default, and almost every libraries are designed for qualified imports, and import their modules qualified. So I'd wager it's the default.

As a software architect, I do actually spend a bunch of my code reviews saying: you should import qualified. It would be a much more effective and powerful message to simply set the default imports as being qualified in my projects. For me, the change in this proposal would really be a very significant change.

Now, the committee may decide that this is still not worth the confusion implied by having two incompatible syntactic conventions out there. That's entirely fair! I just don't want anybody to walk out of this conversation with the feeling that this proposal is an inconsequential stylistic change.

/Arnaud

On Mon, Sep 16, 2019 at 2:04 PM Sandy Maguire <sandy at sandymaguire.me<mailto:sandy at sandymaguire.me>> wrote:
I'm happy with your reasoning, Simon, and am also in favor of rejection.

On Mon, Sep 16, 2019 at 9:23 AM Simon Marlow <marlowsd at gmail.com<mailto:marlowsd at gmail.com>> wrote:
Dear steering committee -

The discussion following my earlier suggestion to reject the proposal has petered out. Taking into account the discussion, it still seems to me that we should reject the proposal, so I've posted on the thread to this effect: https://github.com/ghc-proposals/ghc-proposals/pull/220#issuecomment-531666589

Any further comments before we close it?

Thanks
Simon

On Fri, 5 Jul 2019 at 08:19, Simon Marlow <marlowsd at gmail.com<mailto:marlowsd at gmail.com>> wrote:
Dear steering committee -

I am inclined to reject this proposal, so as per the new committee process I posted the rationale on the github thread: https://github.com/ghc-proposals/ghc-proposals/pull/220#issuecomment-508414602

You may want to consider the proposal and offer opinions while we wait for the authors' rebuttal. It's a very simple proposal.

Cheers
Simon

On Wed, 3 Jul 2019 at 08:55, Joachim Breitner <mail at joachim-breitner.de<mailto:mail at joachim-breitner.de>> wrote:
Dear Committee,

this is your secretary speaking:

QualifiedImports
has been proposed by Arnaud Spiwack and Guillaume Bouchard
https://github.com/ghc-proposals/ghc-proposals/pull/220
https://github.com/tweag/ghc-proposals/blob/qualified-import/proposals/0000-default-qualified-import.rst

I propose Simon M as the shepherd.

Please reach consensus as described in
https://github.com/ghc-proposals/ghc-proposals#committee-process
In particular, talk to the authors before, if you think this should be
rejected, and kick off the discussion on Github, following the steps
described under “Now the shepherd proposes to accept or reject the
proposal” in the above link.

Thanks,
Joachim
--
Joachim Breitner
  mail at joachim-breitner.de<mailto:mail at joachim-breitner.de>
  http://www.joachim-breitner.de/

_______________________________________________
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

--
I'm currently travelling the world, sleeping on people's couches and doing full-time collaboration on Haskell projects. If this seems interesting to you, please consider signing up as a host! https://isovector.github.io/erdos/
_______________________________________________
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/20190917/24186ea4/attachment-0001.html>