Re: [PATCH v5 4/5] c++modules: report imported CMI files as dependencies

[Ben Boeckel <ben.boeckel@kitware.com>]

On Thu, Jun 22, 2023 at 17:21:42 -0400, Jason Merrill wrote:
> On 1/25/23 16:06, Ben Boeckel wrote:
> > They affect the build, so report them via `-MF` mechanisms.
> 
> Why isn't this covered by the existing code in preprocessed_module?

It appears as though it is neutered in patch 3 where
`write_make_modules_deps` is used in `make_write` (or will use that name
in v7 once I finish up testing). This logic cannot be used for p1689
output because it assumes the location and names of CMI files (`.c++m`)
that will be necessary (it is related to the `CXX_IMPORTS +=` GNU
make/libcody extensions that will, e.g., cause `ninja` to choke if it is
read from `-MF` output as it uses "fancier" Makefile syntax than tools
that are not actually `make` are going to be willing to support). This
codepath is the *actual* filename being read at compile time and is
relevant at all times; it may duplicate what `preprocessed_module` sets
up.

I'm also realizing that this is why I need to pass `-fdeps-format=p1689`
when compiling…there may need to be another, more idiomatic, way to
disable this additional syntax usage in `-MF` output.

--Ben