[Ben]
> Getting compilers that actually make the format. Given that one
> implementation outputs it in a released version but needs
> pre-classification (MSVC), one has an extant patch that works for named
> modules and needs plumbing for header units (GCC), and one needs
> plumbing to get even named modules working (Clang), it's definitely not
> ready for any build system to just start scanning today. Maybe
> clang-scan-deps is more ready then Clang itself; I haven't tested it at
> all yet.

Thanks!

When you said
>> Again, compilers should have ways of asking this information. It should
>> ideally be communicated in the format described by P1689

I thought there were bigger issue in terms of just reading that consensus - I didn't realize it was "just a programming exercise" issue 😊

-- Gaby

-----Original Message-----
From: Ben Boeckel <ben.boeckel_at_[hidden]>
Sent: Wednesday, April 20, 2022 1:24 PM
To: Gabriel Dos Reis <gdr_at_microsoft.com>
Cc: sg15_at_[hidden]; Nicolai Josuttis via Ext <ext_at_lists.isocpp.org>; Nicolai Josuttis <nico_at_josuttis.de>; Peter Dimov <pdimov_at_gmail.com>; Tom Honermann <tom_at_honermann.net>; Nathan Sidwell <nathan_at_acm.org>
Subject: Re: [SG15] [isocpp-ext] Can we expect that all C++ source files can have the same suffix?

On Wed, Apr 20, 2022 at 19:12:32 +0000, Gabriel Dos Reis wrote:
> [Ben]
> > Again, compilers should have ways of asking this information. It should
> > ideally be communicated in the format described by P1689
>
> Haven't we (SG15) agree on this already? What else is left?

Getting compilers that actually make the format. Given that one
implementation outputs it in a released version but needs
pre-classification (MSVC), one has an extant patch that works for named
modules and needs plumbing for header units (GCC), and one needs
plumbing to get even named modules working (Clang), it's definitely not
ready for any build system to just start scanning today. Maybe
clang-scan-deps is more ready then Clang itself; I haven't tested it at
all yet.

--Ben