220 38741 <077d3c60-6cc4-42aa-9b18-5dbb593b107b@isocpp.org> article Path: news.gmane.org!.POSTED!not-for-mail From: Charles Salvia Newsgroups: gmane.comp.lang.c++.isocpp.proposals Subject: Re: Proposal for generic rebind template Date: Tue, 19 Jun 2018 18:30:05 -0700 (PDT) Lines: 221 Approved: news@gmane.org Message-ID: <077d3c60-6cc4-42aa-9b18-5dbb593b107b@isocpp.org> References: <824bd50b-55f9-4ca5-b17f-db5ba21245dd@isocpp.org> <7f5b55e9-e935-421a-b0ab-3b5cbbb1c624@isocpp.org> <17d39779-2bb9-4073-ab3f-4b14eef2d0de@isocpp.org> <600a547b-484d-4c2e-812e-cf3d9762ea71@isocpp.org> Reply-To: std-proposals@isocpp.org NNTP-Posting-Host: blaine.gmane.org Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="----=_Part_27087_463653750.1529458205129" X-Trace: blaine.gmane.org 1529458081 27253 195.159.176.226 (20 Jun 2018 01:28:01 GMT) X-Complaints-To: usenet@blaine.gmane.org NNTP-Posting-Date: Wed, 20 Jun 2018 01:28:01 +0000 (UTC) To: ISO C++ Standard - Future Proposals Original-X-From: std-proposals+bncBDY6DLOFSEPBBHW4U3MQKGQEBFQDICQ@isocpp.org Wed Jun 20 03:27:57 2018 Return-path: Envelope-to: gclcip-std-proposals@m.gmane.org Original-Received: from mail-yb0-f198.google.com ([209.85.213.198]) by blaine.gmane.org with esmtp (Exim 4.84_2) (envelope-from ) id 1fVRue-0006yC-HJ for gclcip-std-proposals@m.gmane.org; Wed, 20 Jun 2018 03:27:56 +0200 Original-Received: by mail-yb0-f198.google.com with SMTP id d1-v6sf994350ybq.16 for ; Tue, 19 Jun 2018 18:30:07 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=isocpp-org.20150623.gappssmtp.com; s=20150623; h=date:from:to:message-id:in-reply-to:references:subject:mime-version :x-original-sender:reply-to:precedence:mailing-list:list-id :list-post:list-help:list-archive:list-subscribe:list-unsubscribe; bh=JDf3+nYsHp+kqXQY6QZPsMnfPSWJl+f+CqPTD48p4ZQ=; b=VuJ6sKQ0FUPRa3I97OTyC/kvm3AvvHmU3my7kTQQO0vhRQ9QArfYFZLtZHM6ps48z/ Kl8gAgXjIjcenJTcdNXHEOO1G9e4WAxtHz9OytB5rrAWG9Tterh0xSTcx0R0MrikN0mc BPgBMge8aDBXEPCAHtoICUiE8GQmw9WvjXipZ4y7LC65NLIJT0UXsbRg8Lo5qeXFN8Xa uHzmBXo5vUDRvqVSchknspD/MIjYpzy4ymvlrSiBOkyB6OycY9s1KpnOieLFjibBRBsh fbBaAKcjEB80dc5nxTRJWI/aW/8K0BjnScQS+IcdHzzrAAKA8GSbCNpWwzD9VZIiJYEc F1lg== DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=date:from:to:message-id:in-reply-to:references:subject:mime-version :x-original-sender:reply-to:precedence:mailing-list:list-id :list-post:list-help:list-archive:list-subscribe:list-unsubscribe; bh=JDf3+nYsHp+kqXQY6QZPsMnfPSWJl+f+CqPTD48p4ZQ=; b=t6X4Y0oO5AyNWdwr95wawV815Cw2Y+rRxwfOfuEKJFJq5jgVAN0MxSkNwuEQvgERCa y/bVGoJvpwNGWOVid//MEML+AMDnYt7SHaNr2oIMjRVVKiVFT3pIeCtMsI9U4oLAC1aF tvL4/Rq1k6OB3YoeOFu+xH0BdUHuQYPCE1LJy9ugLdvsf/+VtdEgcXDyUvGtxd6sYppP 16lMMw7/+tG87PdaeTouKJPyLWQSLB17F6+wlxzQ84fHK1/7ZQYWlVPM3+/NhUme42ge WozTz1OzUQof56CiyZR9vx0+oBi3V2/qom44oi7Y6eLw+ICEQopKB5K0BUvRyz3NSNGe srzA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:from:to:message-id:in-reply-to:references :subject:mime-version:x-original-sender:reply-to:precedence :mailing-list:list-id:x-spam-checked-in-group:list-post:list-help :list-archive:list-subscribe:list-unsubscribe; bh=JDf3+nYsHp+kqXQY6QZPsMnfPSWJl+f+CqPTD48p4ZQ=; b=WiSuBiOPFxmwgx1MEN3qVBxtJFIK7ohzHuOsppMfXXfDtbFLRdrr9cDlT0aP83+8M+ H8f1TeDi6J+e5VuUYM1nNbul6xweZnHZalBM6G46qfN5drrRKAqrffR4fcdbnLfQiien JWoLYCGOpeWc4euCi7S4QozF3riuj+CE/dBYuJbq8pEZFkhhdXs15GHeuogELJtyWkmq C6MRrd9zztwdrR41rX3Ni+sMuxLSOUj0CZeKmTz4u7mq6iibTSA/+01zPfATDhdfJOBU 5ff94UHUq+AlSc+YniguHXbARCyonunhelPEtUGz7/SUzyzkCtbo3vb4NZ7V+zcML98F q4BQ== X-Gm-Message-State: APt69E2An4HAKsbIKyvKaxBr036M7OOFBGEeBEbPSEg83zTwiGCajnUR DNKoPijUQcLUTNzyO8COO4IeQQ== X-Google-Smtp-Source: ADUXVKLj5/1ps2CjlqBtaqs0bmPUczYNw/Wj1of9ZjYNaotM89+Z3oMqFgXkYOD/ja9mS0VwAAbVsA== X-Received: by 2002:a81:3e03:: with SMTP id l3-v6mr5828441ywa.60.1529458207351; Tue, 19 Jun 2018 18:30:07 -0700 (PDT) X-BeenThere: std-proposals@isocpp.org Original-Received: by 2002:a81:6e82:: with SMTP id j124-v6ls164182ywc.42.gmail; Tue, 19 Jun 2018 18:30:06 -0700 (PDT) X-Received: by 2002:a0d:dd09:: with SMTP id g9-v6mr947967ywe.6.1529458205736; Tue, 19 Jun 2018 18:30:05 -0700 (PDT) In-Reply-To: <600a547b-484d-4c2e-812e-cf3d9762ea71@isocpp.org> X-Original-Sender: charles.a.salvia@gmail.com Precedence: list Mailing-list: list std-proposals@isocpp.org; contact std-proposals+owners@isocpp.org List-ID: X-Spam-Checked-In-Group: std-proposals@isocpp.org X-Google-Group-Id: 399137483710 List-Post: , List-Help: , List-Archive: List-Subscribe: , List-Unsubscribe: , Xref: news.gmane.org gmane.comp.lang.c++.isocpp.proposals:38741 Archived-At: ------=_Part_27087_463653750.1529458205129 Content-Type: multipart/alternative; boundary="----=_Part_27088_1789772432.1529458205129" ------=_Part_27088_1789772432.1529458205129 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Yeah, I see what you mean. I suppose it's unclear to me actually if that's= =20 bad. The default argument for template parameter 2 in std::vector is=20 std::allocator, not std::pmr::polymorphic_allocator. Default arguments are= =20 really a property of the *template*, not a particular instantiation - so=20 naturally, if you omit the second parameter in the rebind expression,=20 relying on the default parameter, then it would of course have to use the= =20 default parameter used by the *template*, which is std::allocator. So, I= =20 guess you could argue this is to be the expected behavior. You'd need to= =20 explicitly spell out the second argument if you wanted to use a different= =20 allocator type. Regardless, I agree with you that cases like this certainly call into=20 question the overall usefulness of an attempt to write a generic rebind=20 template. On Tuesday, June 19, 2018 at 6:20:54 PM UTC-4, Arthur O'Dwyer wrote: > > On Tuesday, June 19, 2018 at 2:57:36 PM UTC-7, Charles Salvia wrote: >> >> Right. In that case it wouldn't do the expected thing, because=20 >> std::pmr::vector is defined as an alias template rather than a class=20 >> template. >> > > Even if you spell out the type, your "rebind" STILL does the wrong thing. > > using V =3D std::vector>; > using W =3D rebind_t; > static_assert(is_same_v std::pmr::polymorphic_allocator>>); // Oops! > > =E2=80=93Arthur > > =20 > >> On Tuesday, June 19, 2018 at 1:20:10 PM UTC-4, Arthur O'Dwyer wrote: >>> >>> On Monday, June 18, 2018 at 9:03:12 AM UTC-7, Charles Salvia wrote: >>>> >>>> *Updated draft proposal for generic rebind template:* >>>> >>>> http://charles-salvia.github.io/CPP-REBIND-PROPOSAL.html >>>> >>>> Implementation files: >>>> >>>> https://github.com/charles-salvia/std-proposal-generic-template-rebind= =20 >>>> >>>> >>>> This is a proposal to create a generic mechanism for rebinding templat= e=20 >>>> parameters (std::rebind_t), to remove the need for ad-hoc rebind templ= ates,=20 >>>> such as std::allocator_traits::rebind_alloc, etc.=20 >>>> >>> >>> This is a bad idea for several philosophical reasons (and I say that as= =20 >>> the presenter of this talk =20 >>> on a similar subject. The difference is that I don't want any "generic"= =20 >>> rebind facility, I just want a greater number of special-purpose rebind= ers). >>> >>> However, your idea doesn't work anyway. Try it on some library types: >>> >>> using V =3D std::pmr::vector; >>> using W =3D rebind_t; >>> static_assert(std::is_same_v>); //=20 >>> error! >>> >>> Since it's a bad idea *and* doesn't work, I think you should give it up= .. >>> >>> However, it's also much easier to implement as a pure library extension= =20 >>> than you thought it was. I can do it in 28 lines and I bet it can be do= ne=20 >>> in fewer. >>> https://wandbox.org/permlink/pRIQ8lJaf8irznwK >>> >>> =E2=80=93Arthur >>> >> --=20 You received this message because you are subscribed to the Google Groups "= ISO C++ Standard - Future Proposals" group. To unsubscribe from this group and stop receiving emails from it, send an e= mail to std-proposals+unsubscribe@isocpp.org. To post to this group, send email to std-proposals@isocpp.org. To view this discussion on the web visit https://groups.google.com/a/isocpp= ..org/d/msgid/std-proposals/077d3c60-6cc4-42aa-9b18-5dbb593b107b%40isocpp.or= g. ------=_Part_27088_1789772432.1529458205129 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Yeah, I see what you mean.=C2=A0 I suppose it's unclea= r to me actually if that's bad.=C2=A0 The default argument for template= parameter 2 in std::vector is std::allocator, not std::pmr::polymorphic_al= locator.=C2=A0 Default arguments are really a property of the template, not a particular instantiation - so naturally, if you omit the second p= arameter in the rebind expression, relying on the default parameter, then i= t would of course have to use the default parameter used by the *template*,= which is std::allocator.=C2=A0 So, I guess you could argue this is to be t= he expected behavior.=C2=A0 You'd need to explicitly spell out the seco= nd argument if you wanted to use a different allocator type.

<= /div>
Regardless, I agree with you that cases like this certainly call = into question the overall usefulness of an attempt to write a generic rebin= d template.

On Tuesday, June 19, 2018 at 6:20:54 PM = UTC-4, Arthur O'Dwyer wrote:
On Tuesday, June 19, 2018 at 2:57:36 PM UTC-7, Charles S= alvia wrote:
Right.= =C2=A0 In that case it wouldn't do the expected thing, because std::pmr= ::vector is defined as an alias template rather than a class template.
<= /div>

Even if you spell out the type, your = "rebind" STILL does the wrong thing.

=C2= =A0 =C2=A0 using V =3D std::vector<int, std::pmr::polymorphic_alloc= ator<int>>;
=C2=A0 =C2=A0 using W =3D rebind_t<V, dou= ble>;
=C2=A0 =C2=A0 static_assert(is_same_v<W, std::vector&= lt;double, std::pmr::polymorphic_allocator<int>>>); // Oop= s!

=E2=80=93Arthur

=C2=A0=
On Tuesday, = June 19, 2018 at 1:20:10 PM UTC-4, Arthur O'Dwyer wrote:
On Monday, June 18, 2018 at 9:03:= 12 AM UTC-7, Charles Salvia wrote:
Updated draft proposal for generic rebind template:


Implementation files:


This is = a proposal to create a generic mechanism for rebinding template parameters = (std::rebind_t), to remove the need for ad-hoc rebind templates, such as st= d::allocator_traits::rebind_alloc, etc.=C2=A0
=

This is a bad idea for several philosophical reasons (a= nd I say that as the presenter of this talk on a similar subject. The difference is th= at I don't want any "generic" rebind facility, I just want a = greater number of special-purpose rebinders).

Howe= ver, your idea doesn't work anyway. Try it on some library types:
=

=C2=A0 =C2=A0 using V =3D std::pmr::vector<int>;<= /div>
=C2=A0 =C2=A0 using W =3D rebind_t<V, double>;
= =C2=A0 =C2=A0 static_assert(std::is_same_v<W, std::pmr::vector<d= ouble>>); =C2=A0// error!

Since it's a b= ad idea and doesn't work, I think you should give it up.<= /div>

However, it's also much easier to implement as= a pure library extension than you thought it was. I can do it in 28 lines = and I bet it can be done in fewer.

=E2=80=93Arthur
=

--
You received this message because you are subscribed to the Google Groups &= quot;ISO C++ Standard - Future Proposals" group.
To unsubscribe from this group and stop receiving emails from it, send an e= mail to std-proposa= ls+unsubscribe@isocpp.org.
To post to this group, send email to std-proposals@isocpp.org.
To view this discussion on the web visit https://groups.google.= com/a/isocpp.org/d/msgid/std-proposals/077d3c60-6cc4-42aa-9b18-5dbb593b107b= %40isocpp.org.
------=_Part_27088_1789772432.1529458205129-- ------=_Part_27087_463653750.1529458205129-- .