/usr/portsnap vs. /var/db/portsnap

[Scott Long]

M. Warner Losh wrote:
> In message: <42F63353.7030707 at freebsd.org>
>             Colin Percival <cperciva at freebsd.org> writes:
> : M. Warner Losh wrote:
> : > In message: <42F61960.4020400 at freebsd.org>
> : >             Colin Percival <cperciva at FreeBSD.org> writes:
> : > : very little reason for anyone to be running
> : > : a portsnap mirror unless it's a public mirror,
> : > 
> : > Our experience with cvsup would suggest otherwise.  Many places with
> : > large numbers or even small numbers of machines run cvsup mirrors that
> : > are private.  I expect that universities will want to run mirrors that
> : > they might not want non-students accessing (eg, internal bandwidth is
> : > free, external is expensive).
> : 
> : Portsnap != CVSup.  In particular, an HTTP proxy which is used by five
> : hundred users running portsnap will use less bandwidth than a portsnap
> : mirror.  The "right" solution for nearly all organizations is a caching
> : HTTP proxy.
> 
> I'm not worried about bandwidth usage so much as I am about
> availability.  The primary reason I cvsup the CVS tree is so that it
> is always available to me locally and I don't have to depend on my ISP
> having my link up.  Proxie http doesn't help with that at all.
> 
> Warner
> 

That is a very alid developer opinion.  Luckily, our users outnumber
our developers my many orders of magnitude.

Scott