From mboxrd@z Thu Jan 1 00:00:00 1970 From: Lucio De Re To: 9fans@cse.psu.edu Subject: Re: [9fans] /sys/src/9/port/devsd.c Message-ID: <20040103103742.A19051@cackle.proxima.alt.za> References: <20040102185042.B25949@cackle.proxima.alt.za> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: ; from David Presotto on Fri, Jan 02, 2004 at 04:35:44PM -0500 Date: Sat, 3 Jan 2004 10:37:43 +0200 Topicbox-Message-UUID: b19ffa48-eacc-11e9-9e20-41e7f4b1d025 On Fri, Jan 02, 2004 at 04:35:44PM -0500, David Presotto wrote: > > They're in pc/dat.h on sources. You're out of date. Yep, I have what I hope is a well-maintained copy of sources and I was quite successful in compiling the kernel with that set. Of course, it is now no longer pristine, is it? It concerns me that there is no auditing mechanism, where one really needs one. My problem is compounded by the line speed, which is 9600bps out of my office (it gets better further along, but I can't always exploit the remote proxy services). You an imagine that a 200Meg check of consistency could take a long time. I'm thinking along the lines of setting up a log/db (I haven't found the need to figure out which does what, I guess I need to do that now) of locally modified files so they can be tucked away during a full (-s) update, then restored with a separate replica/pull. This would enable me (notwithstanding my speed restrictions, I could use a CD image, say) to perform a clean re-install (how will that affect the dump, Fossil/Venti or traditional?) in a live environment. The other option I have thought a little about is not only to have a pristine copy of sources as I do presently, but also to rebuild a parallel copy from the CD and run consistency checks and audits across these. I mention these so anyone who's had similar thinking can make suggestions, I'm still trying to figure out what replication should really be capable of. Hm, I wonder how much one can learn from comparing log/db files with sourcesdump (my CVS roots are showing, I guess)? Could one not add a release file to each update so that it is easier to establish which dump date to consult? Annotating each update may be asking too much, that I concede. ++L