Gentoo Archives: gentoo-portage-dev

From: Scott Taylor <security@××××××××××××××.com>
To: gentoo-portage-dev@l.g.o
Subject: Re: [gentoo-portage-dev] Feature request
Date: Thu, 01 Jan 2004 21:43:39
Message-Id: 1072993399.11196.21.camel@Star.BerthoudWireless.net
In Reply to: Re: [gentoo-portage-dev] Feature request by Paul de Vrieze
1 The default list of sourceforge (and other) mirrors would seem to me to
2 be well suited to being in make.globals where their value and syntax are
3 easily found, and new values can easily be placed in make.conf for those
4 who prefer other mirror sites. and bad mirrors can get yanked. Like for
5 example, every once in a while a gnukorea (or something like that) comes
6 up as an attempted mirror, however an emerge gets stuck there because
7 its asking for a password. If the lists of available mirrors were more
8 easily found and updated, it would be a lot easier to fix things like
9 that.
10
11 And I wouldn't expect it to be too difficult to add a
12 FEATURE="ignorenomirror". The machines inside my network only try my
13 local mirror, plus one of the gentoo mirrors before trying the
14 non-mirrored source uri anyway, so ignoring the nomirror directive is
15 not going to cause a drastic increase in file not found requests from
16 the mirrors. The server hosting the distfiles here would be left with a
17 normal set of source mirrors, and obeying the nomirror directive like
18 everyone else.
19
20 I might be the only one trying to do it just like this, but I'm
21 certainly not the only one trying to share out my own distfiles
22 directory on an internal network. I've even considered burning distfiles
23 snapshots on dvd-r to leave in otherwise unused drives. Also, I'd heard
24 a request from catalyst that portage be able to understand directory
25 paths as a mirror source. Thats another fine example of using a local
26 distfile store of some sort that you would much prefer to have it try
27 first, while ignoring a nomirror directive.
28
29
30 On Thu, 2004-01-01 at 14:19, Paul de Vrieze wrote:
31 > This does not mean that mirror selection can and should be done better,
32 > however it is hard to come up with a real good algorithm that does not
33 > involve people changing things by themselves (but still allows them to)
34 >
35 > Paul
36 --
37 Scott Taylor - <security@××××××××××××××.com>
38
39 The Golden Rule of Arts and Sciences:
40 He who has the gold makes the rules.
41
42
43
44
45 --
46 gentoo-portage-dev@g.o mailing list