[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Configuration (File) System



On Thu, 2004-03-18 at 18:36, Oliver Kurth wrote:
> On Thu, 2004-03-18 at 09:40, Jean Pierre LeJacq wrote:
> > On Thu, 18 Mar 2004, Oliver Kurth wrote:
> > 
> > > On Thu, 2004-03-18 at 01:48, Jens Hermans wrote:
> > > > Currently I'm working on a technology to be able to represent
> > > > configuration files as filesystems. The whole thing is outlined at
> > > > http://users.pandora.be/paranet/CFS.htm , but I'll try to sumarize it
> > >
> > > I have a very similar idea, see
> > > http://advogato.org/person/oku/diary.html?start=4
> > 
> > Have you considered the possibility XCAP or ACAP both of which are
> > IETF draft standards.  They seem very close to what you are looking
> > for.  A summary is available at:
> > 
> >   http://xml.coverpages.org/ni2003-05-29-a.html
> 
> Well, Jens explains on his page why he does not want to use xml. I do
> not want either, though it would be possible to add an xml interface,
> for both input and output - once there is a tree, you can make xml out
> of it easily.
> 
> > Along the same lines, wouldn't ldap be a more generic solution than
> > a new file system?
> 
> Same arguments would apply to ldap as for xml.
> 
> At least my goal (cannot speak for Jens), was to have it _simple_, with
> as low overhead as possible, because I am also thinking about embedded
> systems. Something that can be used by package scripts, without them
> having too many dependencies.
> 
> Greetings,
> Oliver

I must agree with Kurth, keep it simple, everyone can use filesystems.
Remember that other programmers will use cfs and that they probably
don't want to spend a lot of time on learning new stuff.

I've quickly read the document about xcap, and it involves an xml
repository, accessible through http (or just as files if local). Maybe
this is good in large systems/clusters, but it's definetly overkill for
the average home system. If this thing should become an industry
standard, an interface might be added to cfs for this, it's not that
difficult i suppose to export the existing fs to xml instead of to the
fuse kernel module. Atm i prefer using the simple method, since it'll
probably get more acceptance than a rather difficult system like xcap
(compared to fs's). 





Reply to: