[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: global .EXTRA_PREREQS not quite right...
From: |
Britton Kerin |
Subject: |
Re: global .EXTRA_PREREQS not quite right... |
Date: |
Tue, 15 Jun 2021 16:16:10 -0800 |
On Tue, Jun 15, 2021 at 4:22 AM Mark Piffer <mark.piffer@gmail.com> wrote:
>
> My solution was to separate the build system into a (non-recursive) structure
> of make files which each have roughly speaking only one recipe. You may raise
> an eyebrow but I can assure you that the segregation of a build process in
> this way works, at least for embedded development. Complicated dependency
> trees with very diverse recipes are bundled into the few non-single-recipe
> makefiles, which the master makefile easily tolerates. The .EXTRA_PREREQS are
> needed only for housekeeping purposes, i.e. creating dependencies on
> makefiles, process scripts and auto-dependencies without them getting into
> the automated input/output set
> of the build process where only real engineering work products are kept.
Sounds reasonable. Regarding the use cases for global .EXTRA_PREREQS
that a previous poster asked about, this is one: it lets you state in
one place that any change to the build rules themselves should trigger
rebuild of everything, without having to paste Makefile $(wildcard
*.mk) serialized_vars etc. everywhere. For my case at least it turned
out there were only a couple places where 'umbrella' targets needed to
be expressed as lists instead, so that's what I've done.
Britton