Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 10 Apr 2014 10:43:13 -0600
From:      Warner Losh <imp@bsdimp.com>
To:        =?windows-1252?Q?Dag-Erling_Sm=F8rgrav?= <des@des.no>
Cc:        arch@freebsd.org
Subject:   Re: ar and ranlib -D
Message-ID:  <257F4CC6-78AC-4729-B0F7-50FDA296D46D@bsdimp.com>
In-Reply-To: <86a9btuqh1.fsf@nine.des.no>
References:  <86eh15usv2.fsf@nine.des.no> <79CBA7AC-998E-46EE-8F94-F92C7C00FF75@bsdimp.com> <86a9btuqh1.fsf@nine.des.no>

index | next in thread | previous in thread | raw e-mail


On Apr 10, 2014, at 9:57 AM, Dag-Erling Smørgrav <des@des.no> wrote:

> Warner Losh <imp@bsdimp.com> writes:
>> My only concern is with the %POSIX section. That change isn’t needed
>> for reproducible builds.
> 
> Is it harmful?
> 
> (what is that, anyway?  It's not documented in make(1))

Short answer: It isn’t defined by POSIX 1003.2, so yes.

POSIX mode, which is barely documented in the .POSIX target, causes
make(1) to try hard to comply with POSIX requirements.  Make, itself, winds
up setting %POSIX to “1003.2” and not remaking the Makefiles.  The global
sys.mk system responds to this variable by only using commands defined
by POSIX 1003.2, so it uses c89, instead of cc. It adds the dash to the ar
command, and a bunch of other silly differences that are none-the-less
mandated by POSIX. Since -D isn’t defined by POSIX ar, your change
breaks that and so is harmful...

Warner



home | help

Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?257F4CC6-78AC-4729-B0F7-50FDA296D46D>