[PATCH 1/7] newlib: libm: merge machine/ trampoline up a level

[Mike Frysinger]

On 28 Jan 2022 14:28, Jon Turney wrote:
> There might be some scope in [2/3] for using ${srcdir} relative paths, 
> rather than making everything absolute?

as it stands atm, i don't think so.  srcdir is not guaranteed to be relative,
it is just often so.  that means you can't assume it is and then try to prefix
it to make it work in subdirs.

so as long as the Makefile is not in the same directory as the configure, the
only reliable thing is to use absolute paths in the Makefile.

that said, some of the work i've done to get rid of recursive make would help
here.  if the subdir doesn't have a Makefile, but has a makefile fragment that
is included in the Makefile that lives alongside the configure script, then we
can use relative paths again.  tbh, i'm not sure how much i'm going to do in
that space.  automake generates a lot faster than autoconf, and recursive make
is faster than recursive configure.  still not ideal, but not the worst.  maybe
i'll finish libm and leave it at that.  at least give time for the codebase to
settle and for people to report regressions.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/newlib/attachments/20220128/39101e69/attachment.sig>