On Tuesday 18 December 2007 06:14:24 pm Chuck Cole wrote: > > -----Original Message----- > > From: Josh Paetzel [mailto:josh at tcbug.org] > > Sent: Tuesday, December 18, 2007 4:36 PM > > To: tclug-list at mn-linux.org > > Cc: Chuck Cole; Dan Rue; Mike Miller > > Subject: Re: [tclug-list] FreeBSD coherence > > > > > Seem both feasible and desirable to make some sort of > > > > after-the-fact "rule > > > > > checker" for Linux to self-enforce such a discipline and detect when > > > something new is contrary. Might need some heuristics in any case, but > > > that tool would preserve the option to take exception as well > > > > as the check > > > > > for anomalies. I think I'd prefer the rule checker since scope and > > > completeness become verifiable and not just an "implied mystique" of > > > the OS. > > > > > > > > > Chuck > > > > Are you suggesting that it's easier to have a tool that tells you > > things are > > in the 'wrong place' than to just put them in the right place in > > the first > > place? That seems counter-intuitive to me, it's always less work > > and more > > robust to do it right the first time than to do it wrong and have a > > second-pass try and fix it. > > One must believe himself sufficiently omniscient for that. I'm still too > low on the learning and retention curve. > > > There is no 'implied mystique' in the FreeBSD ports tree. > > LOCALBASE is set > > to /usr/local, everything in the ports tree defaults to > > installing under that > > hierarchy. A port *can* override LOCALBASE, and in very rare > > occassions it's > > permitted....for instance, if the port installs a kernel module > > of some sort > > it has to be able to make it's way in to the root > > filesystem....but the vast > > majority of the 18,000+ apps in the ports tree simply do not put anything > > outside of /usr/local and it manages to happen without having to somehow > > clean up after the fact. > > I'm way too much of a newbie to have such clairvoyance and/or experience > with this, and too much a skeptic to believe that human or other error > can't creep in before, during, or after a mod. > > For example, I still prefer to have my hard drive do its own error > detection and correction, and gave up doing integrity checks by hand > several weeks ago > > :-) > > Which part of "verification" am I missing? :-) > > > Chuck In the case of the FreeBSD ports tree the system is set up to use LOCALBASE unless specifically molested to do otherwise. I suppose someone "could" commit a port that has LOCALBASE overridden when it shouldn't be, but there's only a few dozen ports committers and they are generally on the lookout for stuff like that. I'd agree that if every single port had to be set up to do the right thing then there would be concern for breakage, but the framework does the right thing by default. The analogy of checking your hard drive in this case is more along the lines of 'creating a system to make sure my hard drive isn't writing to the floppy'....without fairly unusual intervention it's just really not possible. Another possibility is that someone could break the framework itself somehow, but so far in the 12 years I've been using and contributing to FreeBSD no one's broken the ports tree in a way that it leaked files out of LOCALBASE. -- Thanks, Josh Paetzel PGP: 8A48 EF36 5E9F 4EDA 5A8C 11B4 26F9 01F1 27AF AECB -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 187 bytes Desc: This is a digitally signed message part. Url : http://mailman.mn-linux.org/pipermail/tclug-list/attachments/20071218/f03585e4/attachment.pgp