Date: Wed, 09 Apr 2014 13:56:27 +0200 From: John Marino <freebsd.contact@marino.st> To: Big Lebowski <spankthespam@gmail.com>, marino@freebsd.org Cc: "Mikhail T." <mi+thun@aldan.algebra.com>, Tijl Coosemans <tijl@coosemans.org>, freebsd-ports <freebsd-ports@freebsd.org> Subject: Re: FreeBSD ports which are currently scheduled for deletion Message-ID: <5345356B.9080508@marino.st> In-Reply-To: <CAHcXP%2BctBUSmcN2Mcg=7kn3obnqUjgkq-oDPeEB8au48yU9QEA@mail.gmail.com> References: <mailman.0.1396958400.6606.freebsd-ports@freebsd.org> <5344005C.4030503@aldan.algebra.com> <20140408185537.69d5cd6e@kalimero.tijl.coosemans.org> <53442E10.6060907@aldan.algebra.com> <20140409002033.5a2d9850@kalimero.tijl.coosemans.org> <CAHcXP%2BfJUMkK3qPdS_wr99B-xXJxzAOkQ%2Bek=R2aKxUTTE_uCg@mail.gmail.com> <534512CB.5030109@marino.st> <CAHcXP%2BctBUSmcN2Mcg=7kn3obnqUjgkq-oDPeEB8au48yU9QEA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 4/9/2014 13:45, Big Lebowski wrote: > On Wed, Apr 9, 2014 at 11:28 AM, John Marino <freebsd.contact@marino.st>wrote: > >> On 4/9/2014 11:22, Big Lebowski wrote: >>> While we are not having any way to measure ports usage (or am I wrong >>> here?), we're still building packages from ports, and I would hope that >> we >>> could get some statistics of pkg usage for certain packages from official >>> repositories, could we? This is not exactly direct relationship for the >>> port being used but I would think that such knowledge is better than >> none. >> >> How do you distinguish packages downloaded by mirrors versus those >> downloaded by pkg? At DragonFly, we'd love to know how to do this >> because it always comes up when the "it's kill to kill i386 platform" >> discussion comes up. Every time somebody brings up a statistic about >> how many times packages are download (or what % packages downloaded are >> i386) then the very next questions is: are those legitimate downloads. >> > > I dont know, but the numbers of package downloads by mirrors should be the > same for all packages, since mirrors should (in theory) fetch entire > repository, effectively 'nullyfying' their own downloads from the > statistics, and what would stand out, should be actual downloads per > package. I don't think theory works. You're only considering official FreeBSD mirrors and not spiders and scrapers. One really needs positive identification and not raw numbers. And of course, the bias the other direction is a large number of individual machines that subscribe to a non-freebsd mirror such as an NFS drive (one download, multiple uses) > I have already said this is by no means a perfect solution, but having some > data, even if it might be biased more or less seems to be a better thing, > than having no data at all. I'm not sure I buy the idea that bad data is better than no data myself. John
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5345356B.9080508>