From owner-svn-ports-all@FreeBSD.ORG Sat Aug 30 20:38:46 2014 Return-Path: Delivered-To: svn-ports-all@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 910F4FE; Sat, 30 Aug 2014 20:38:46 +0000 (UTC) Received: from f10.opsec.eu (f10.opsec.eu [IPv6:2001:14f8:200:4::2]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 48BA8175B; Sat, 30 Aug 2014 20:38:46 +0000 (UTC) Received: from pi by f10.opsec.eu with local (Exim 4.84 (FreeBSD)) (envelope-from ) id 1XNpQ1-000JQL-HR; Sat, 30 Aug 2014 22:38:41 +0200 Date: Sat, 30 Aug 2014 22:38:41 +0200 From: Kurt Jaeger To: Sunpoet Po-Chuan Hsieh Subject: Re: svn commit: r366405 - in head/databases: . p5-Net-Async-CassandraCQL Message-ID: <20140830203841.GR42528@f10.opsec.eu> References: <201408281407.s7SE77sA063958@svn.freebsd.org> <20140829184115.GL42528@f10.opsec.eu> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Cc: "svn-ports-head@freebsd.org" , "svn-ports-all@freebsd.org" , Andrej Zverev , Kurt Jaeger , "ports-committers@freebsd.org" X-BeenThere: svn-ports-all@freebsd.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: SVN commit messages for the ports tree List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 30 Aug 2014 20:38:46 -0000 Hi! > > > In real life, modules from Test:: namespace usually requires only for > > > tests. No issue for people who using binary packages, but extra deps for > > > people who still using ports. [...] > > So I assume BUILD_DEPENDS is correct. Do you see other options > > on how to handle this ? First, thanks for the many hints on how to handle this. > TEST_DEPENDS is correct. I've prepared a diff at http://people.freebsd.org/~pi/misc/cascql.svndiff When I make test the TEST_DEPENDS are not installed: [...] t/21prepare.t ............... Can't locate Test/Identity.pm in @INC [...] How is TEST_DEPENDS supposed to be used so that this works for the 'make test' case, but not the 'make install' case ? > The author should use "test_requires" instead of "build_requires" in this > case. So an upstream PR is required ? Or is it only recommended ? > The following results show that Test::{Identity,Refcount} are only for test > purpose and not required for runtime. > > % grep -r Test::Identity * [...] Ok, I'll add this to my checks. > For p5-* ports, please make sure the modules listed in Build.PL or > Makefile.PL are truly required. > At least a simply "grep" would help to determine if TEST_DEPENDS is more > suitable, especially for Test::* dependencies. > > BTW, there are something can be improved in the Makefile: > - MASTER_SITE_SUBDIR is unnecessary. > - Please respect minimal version requirement in Build.PL / Makefile.PL, e.g. > Future: '0.17' > IO::Async::Loop: '0.54' > IO::Async::Stream: '0.59' > Protocol::CassandraCQL: '0.11' > - > "devel/p5-Devel-GlobalDestruction>0:${PORTSDIR}/devel/p5-Devel-GlobalDestruction" > is incorrect Those are fixed in the svndiff above. I'll commit it if no further changes are required. I'm still unsure about the TEST_DEPENDS issue. -- pi@FreeBSD.org +49 171 3101372 6 years to go !