Date: Wed, 30 May 2012 15:37:51 +0300 From: Konstantin Belousov <kostikbel@gmail.com> To: David Chisnall <theraven@theravensnest.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org, Marius Strobl <marius@alchemy.franken.de> Subject: Re: svn commit: r236137 - head/contrib/gcc/config/i386 Message-ID: <20120530123751.GL2358@deviant.kiev.zoral.com.ua> In-Reply-To: <15CED26F-127B-4736-9E96-6315D6303B31@theravensnest.org> References: <201205270527.q4R5Rm44028055@svn.freebsd.org> <20120528190355.GA42283@alchemy.franken.de> <20120528204728.GD2358@deviant.kiev.zoral.com.ua> <20120529224833.GW90133@alchemy.franken.de> <20120530034747.GJ2358@deviant.kiev.zoral.com.ua> <20120530080151.GX90133@alchemy.franken.de> <15CED26F-127B-4736-9E96-6315D6303B31@theravensnest.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--1PHFXgj3K5uCxEyX Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Wed, May 30, 2012 at 10:00:02AM +0100, David Chisnall wrote: > On 30 May 2012, at 09:01, Marius Strobl wrote: >=20 > > Ehm, yes, but given that this wouldn't be the first such flag we have > > is avoiding it really worth the link time and run time overheads in > > the long term?=20 >=20 > Given the small overhead of the extra hashes, yes. At some point in > the future, we can turn off the older ones and get a tiny reduction > in overhead, but doing it now would cause much more pain for users in > not being able to copy binaries from slightly newer to slightly older > machines than we'd save from a tiny increase in binary size. > > This is the archetypal change for incremental deployment, let's not > make our users' lives difficult just because we can. > > David Who doesn't want to be woken up by mobs of users with flaming > torches and pitchforks. I agree, the overhead of additional symbol hash is relatively low, and only impact developers. I do not think that it is reasonable to even start thinking about turning off SysV hash before at least one release passes. Besides this, another consideration to keep SysV hash around is to keep other tools, besides rtld, working. I am not sure whether e.g. libelf or DTrace use hash. But if using hash, I am quite sure that they use SysV one. --1PHFXgj3K5uCxEyX Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (FreeBSD) iEYEARECAAYFAk/GFJ8ACgkQC3+MBN1Mb4jwRwCggQF0Xt7xoj5yphYJ44tHkrld okQAmgP++hLI0VcmrvM2aQPGeZNH1mnf =Xsdo -----END PGP SIGNATURE----- --1PHFXgj3K5uCxEyX--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120530123751.GL2358>