Steffen Nurpmeso <stef...@sdaoden.eu> wrote:
> Hello Omar.
>
> Omar Polo wrote in
> <28P1AX08SNCS0.30TJDZ4GTJ8IJ@venera>:
> |Steffen Nurpmeso <stef...@sdaoden.eu> wrote:
> |> Stuart Henderson wrote in
> |> <yn2oicg+bo0dc...@bamboo.spacehopper.org>:
> |>|On 2022/05/13 00:22, Steffen Nurpmeso wrote:
> |>|> Since the S-nail port did not honour port CFLAGS for i think
> |>|> years, i hope it is ok to go like this for some time.
> |>|
> |>|That was because nobody had noticed it. Ports policy is to honour CFLAGS.
> |>
> |> Sigh. Works just fine on CRUX and Alpine Linux ;)
> |> Well i do not know. The Python Makefile.inc does a virtuous
> |> dance, CRUX does "make EXTRA_CFLAGS="$CFLAGS"" after doing so (for
> ...
> |well, I can't speak for crux nor alpine, but as a general rule we want
> |to be in control of the CFLAGS the port uses when building. Sometimes
> |one port slips, but that's the policy.
> |
> |If a compiler changes something (see for example the not too old clang
> |issue regarding -fcommon) you want to be able to fix the fallout by
> |tweaking the various Makefiles, not digging into every port build system
> |and patching things.
>
> OK i understand the desire to add additional flags per se.
(i forgot to add that this mechanisms is also used to generated debug
packages -- see bsd.port.mk for "THE DEBUG_PACKAGES INFRASTRUCTURE --
and possibly other things I haven't thought yet. Oh, and from the POV
of a project, it's better to *append*, not prepend, the CFLAGS passed,
so users and packagers can effectively control the build process from
the outside)
> |> Adding onto the environment's $CFLAGS the set of project's CFLAGS
> |> is ok? I see what i can do tomorrow.
> |
> |in this case it's as simple as (from the port makefile)
> |
> | MAKE_FLAGS += CFLAGS="${CFLAGS}"
>
> If you would be so kind to add at least -DNDEBUG i would be fine.
> In fact i wanted to go the hard way and add an EXTRA_CFLAGS
> mechanism, and patch it into the port, but for OpenBSD it is ok.
> Note please also do LDFLAGS="${LDFLAGS}" then, because the
> built-in ones use -pie, and that requires -fPIE in CFLAGS, which
> we then cannot guarantee.
imported, thanks! (and audio/s-cdda-to-db too)
I didn't add -fPIE because on OpenBSD clang generates PIE code by
default (see clang-local(1) for the OpenBSD-specific behavior of llvm)
and I didn't add -DNDEBUG because I don't see any calls to assert(3).
if I've missed something please don't hesitate to point it out so that
we can fix it!
regarding the other port, s-cdda-to-db, I've added NO_BUILD and PKG_ARCH
and tweaked a bit pkg/DESCR to avoid the post scriptum and to fix a typo
in "informations".
> A possible next release of s-cdda will have an EXTRA_CFLAGS
> mechanism, already committed.
>
> |and if you're ok with that and sthen@ doesn't change his mind i'll
> |import it.
>
> Thank you very much.
>
> --steffen
> |
> |Der Kragenbaer, The moon bear,
> |der holt sich munter he cheerfully and one by one
> |einen nach dem anderen runter wa.ks himself off
> |(By Robert Gernhardt)
