9fans - fans of the OS Plan 9 from Bell Labs
 help / color / mirror / Atom feed
From: "Russ Cox" <rsc@plan9.bell-labs.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] replica
Date: Wed, 29 May 2002 12:32:34 -0400	[thread overview]
Message-ID: <468e18b953b35e090c738bfc9994928e@plan9.bell-labs.com> (raw)

> the advantage of keeping a hash (eg, md5 or sha1) is that it distinguishes
> contents of files, which usually is what really matters,
> whereas modification times and file length by themselves
> aren't really adequate (in general, in my experience).

> in short, i thought the use of hash values in wrap was a good idea
> and was surprised it wasn't part of replica!

i may well have dropped the ball on this one.
i'm still not sure.  replica was written only
for the distribution and as such i had little
experience with it before sending it into battle.
originally i used only mtimes and thought that
having the local metadata kept by the file system
was a feature, so you could play games if you knew
what you were doing.  then i got talked into using lengths
too (and rightfully so), and at that point i should
probably have tossed in the hash.
at this point it's hard to add fields to the database
and the log, since the current replica tools won't
like that.

i originally planned to use tra but it was clear that
tra wouldn't be solid enough by the time it was
needed.  once i understand things better i may set
up tra as an alternate way to get updates.  what i like
best about tra is that it's completely general.

with replica you have to declare a server to be the single
point of truth, and if you try to chain things (e.g.,
you try to grab updates using a different machine as
server, or vita nuova adds or tweaks some files) then
you have to be careful, because in general you can't
just grab updates from any other machine and expect everything
to work.

if we were using tra, you could install a new system
using any other plan 9 system you had available on the
network, and it would still be safe to go back to
sources for updates.  i still want to see what that
would be like, but it won't happen until the fall.

russ


             reply	other threads:[~2002-05-29 16:32 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-05-29 16:32 Russ Cox [this message]
  -- strict thread matches above, loose matches on Subject: below --
2006-08-25  2:30 Richard Bilson
2006-08-25  3:38 ` Lyndon Nerenberg
2006-08-25  4:08   ` geoff
2006-08-25  8:19     ` Martin Neubauer
2006-08-27 23:23       ` geoff
2006-08-28  2:46         ` Richard Bilson
2005-02-12 13:42 Sergey Reva
2005-02-12 16:15 ` Russ Cox
2004-03-24 13:20 David Presotto
2002-07-30 12:15 Russ Cox
2002-07-04 10:20 Fco.J.Ballesteros
2002-07-04 10:16 okamoto
2002-07-04  9:26 Fco.J.Ballesteros
2002-07-04  9:21 okamoto
2002-05-29  9:43 forsyth

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=468e18b953b35e090c738bfc9994928e@plan9.bell-labs.com \
    --to=rsc@plan9.bell-labs.com \
    --cc=9fans@cse.psu.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).