Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 11 Nov 2003 04:04:48 +0000
From:      Mark Linimon <linimon@lonesome.com>
To:        Fernan Aguero <fernan@iib.unsam.edu.ar>
Cc:        freebsd-ports@freebsd.org
Subject:   Re: Ability for maintainers to update own ports
Message-ID:  <3FB05FE0.2090005@lonesome.com>
In-Reply-To: <20031111144221.GA527@iib.unsam.edu.ar>
References:  <1068458390.38101.19.camel@dirk.no.domain> <20031110152000.622db381.lehmann@ans-netz.de> <1068471598.38101.77.camel@dirk.no.domain> <20031110163623.GC93583@procyon.firepipe.net> <3FB02895.5050108@ciam.ru> <20031111001932.GA95315@toxic.magnesium.net> <20031111144221.GA527@iib.unsam.edu.ar>

next in thread | previous in thread | raw e-mail | index | archive | help
>
>
>what is the potential impact of doing a
>less-than-perfect port? Breaking hier(7)?
>
Due to the fact that all the ports stuff is based on Makefiles
(arguably our Achilles' Heel), it's actually possible to break a
large number of ports at once.  You've no doubt seen these
annoying "index build failed" messages -- that indicates that
because someone made an error in a ports Makefile, it is
no longer possible to build a spanning tree of ports dependencies,
and (IIIRC) portupgrade and other code relies on that
spanning tree to be correct.

So, it isn't just simple "resistance to change" here, there's
techical reasons, too.

As for the tempting next idea, "let's get rid of the Makefile-
based technology", well ... the QA that would be needed
to test the nearly 10,000 ports would be daunting, to say
the least.  Not to mention the conversion effort.

There are a couple of ideas floating around on the net to
do applications via some kind of metadescription, but those
solutions are months if not years from being reality.

In the meantime, it really comes down to just getting more
volunteers involved within the existing framework, IMHO.

mcl





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3FB05FE0.2090005>