Re: pkg/48500: lang/perl5 cause an unaligned access exception.

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

The following reply was made to PR pkg/48500; it has been noted by GNATS.

From: Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>
To: eric%cirr.com@localhost
Cc: gnats-bugs%NetBSD.org@localhost, pkgsrc-bugs%netbsd.org@localhost, 
tsutsui%ceres.dti.ne.jp@localhost
Subject: Re: pkg/48500: lang/perl5 cause an unaligned access exception.
Date: Sun, 26 Jan 2014 22:51:03 +0900

 > nullnilaki%gmail.com@localhost writes:
 > - >Fix:
 > - Please apply this patch at pkgsrc/lang/perl5/hacks.mk
 > - ### gcc-4.*.* in NetBSD/alpha causes  unaligned access exception in perl.
 > - ### -O works around, and there is a report that -O2 -fno-tree-ter is 
 > enough.
 > - .if !empty(MACHINE_PLATFORM:MNetBSD-*-alpha) && 
 > !empty(CC_VERSION:Mgcc-4.*.*)
 > - # XXX: is there any good way to replace the default -O2 with multiple args?
 > - #PKG_HACKS+=            optimisation
 > - #BUILDLINK_TRANSFORM+=  rename:-O[2-9]*:-O2 -fno-tree-ter
 > - CFLAGS+=-O2 -fno-tree-ter
 > - .endif
 > 
 > I'm integrating this into the tree, but I can't decide if I need
 > to bump PKGREVISION on perl or not.  The issue only affects
 > NetBSD/alpha (or maybe alpha's generically), and otherwise has
 > no effect.
 > 
 > And I'd like to avoid a huge churn on non-alpha platforms.
 > 
 > My inclination is not to bump it, as fixes an issue (poor
 > execution style) on one platform only.
 > 
 > Contrary opinions?
 
 Simply bump PKGREVISION per principle.
 
 As you see, discussion for a human judgment per feelings
 without a responsible person just wastes much more time
 than automated binary rebuilds, due to random bikeshed comments.
 
 ---
 Izumi Tsutsui