Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 12 Feb 2009 05:14:22 -0600
From:      linimon@lonesome.com (Mark Linimon)
To:        Gerald Pfeifer <gerald@pfeifer.com>
Cc:        ports@FreeBSD.org, QAT@FreeBSD.org, ports-committers@FreeBSD.org
Subject:   Re: bitrot [was: cvs commit: ports/science/mbdyn Makefile]
Message-ID:  <20090212111422.GB19851@soaustin.net>
In-Reply-To: <alpine.LSU.1.99.0902071738370.6168@acrux.dbai.tuwien.ac.at>
References:  <200902071511.n17FBVqH073928@repoman.freebsd.org> <20090207160151.829C78FC71@release.ixsystems.com> <alpine.LSU.1.99.0902071738370.6168@acrux.dbai.tuwien.ac.at>

next in thread | previous in thread | raw e-mail | index | archive | help
[Cc: trimmed]

On Sat, Feb 07, 2009 at 05:44:13PM +0100, Gerald Pfeifer wrote:
> We really shouldn't have so many ports without a real maintainer properly
> taking care.
> 
> The amount of bitrot in the tree is worrisome [...]

Agreed; however, in the past the question has been, "who decides what
ports should be removed"?

I once attempted to do a survey based on "last time port was updated";
however, it did not result in what I expected.  In particular, some of
the ports in math/ had not been updated for a while -- but appeared to
still be not only the latest version, but also still useful.  One might
assume that certain ports are "good enough" as they are.

This is by no means to deny there's bitrot.  There's a continuum
in terms of "serves its purpose correctly" to "needs to be updated
continuously" (for the latter, imagine web browsers).  We have portscout
to at least help us a bit in the update category -- and I am trying
to see what has broken in the distfile survey to try to fix that end.
Of course, various portmgrs work on setting things broken/to-expire,
and so forth, and we send out emails based on that.

But there's no good metric currently to determine "is anyone using
this port or not", other than to mark it for deprecation, push the
email about deprecated ports, and see who reacts.  (FreshPorts
subscriptions and sysutils/bsdstats output are opt-in, and thus
probably not statistically significant).

So, I think this is an open topic.

mcl



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090212111422.GB19851>