Date: Mon, 9 May 2005 10:43:38 -0500 From: Greg 'groggy' Lehey <grog@FreeBSD.org> To: Jean-Sebastien Pedron <jspedron@club-internet.fr> Cc: freebsd-fs@freebsd.org Subject: Re: Call for feedback about ReiserFS import Message-ID: <20050509154338.GD1592@eucla.lemis.com> In-Reply-To: <427F491C.4090501@club-internet.fr> References: <427F491C.4090501@club-internet.fr>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, 9 May 2005 at 13:27:24 +0200, Jean-Sbastien Pdron wrote: > Hi everyone, > > The current port of ReiserFS is ready to be imported in the tree for a > few months, however my mentor mux@ and I have concerns about the > organization of the sources of this version. > > Originally, the port is based on Linux 2.6.7. My goal was to distribute > it as a 3rd party package, maybe a port. Because I was interested in > kernel development and how ReiserFS was working, I started from scratch, > bringing the necessary pieces of code from Linux little by little, to > achieve read-only support. > > The main issue with this approach is maintainability: the filenames and > source layout is different between the original Linux version and the > port. Changes in the Linux tree are harder to track. > > I'd like to import this version in FreeBSD's tree, because: > - ReiserFS' homepage (http://www.namesys.com/) says it shouldn't evolve > anymore (just rare bugfixes). For instance, between Linux 2.6.7 and > 2.6.11.3, they added a few checks of functions' return code. > - with my approach, I have a better knowledge of the filesystem > internals. If we want a BSD licensed version, it should be easier. > - work is more interesting this way than just make the Linux version > compile. This may be true, but it isn't really relevant to what goes into the tree. > - it's already done ;) Well, as far as it goes. I wouldn't say it's done until you have the complete functionality. > But I know that the tradition is to import external code with changes as > less intrusive as possible, for good reason. Does anyone has objections > for this? I don't know if "objection" is the right word, but it seems to be a sub-optimal approach. Also, there's a real danger that the full functionality might never be achieved. On the other hand, it's better than nothing, and since there's also a real danger that nobody will do a full version anyway, I'm not going to take a position. Greg -- See complete headers for address and phone numbers
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050509154338.GD1592>