- From: carmen <_@whats-your.name>
- Date: Fri, 29 Aug 2014 22:10:53 +0000
- To: public-rww@w3.org
> someone rearranges their cloud storage server? Are files still discoverable? no, everything breaks. at least on the servers i tend to use. link-rot is as easy as mv(1).. but there's links all the way down. multicians made their file-system a directed-graph long ago: http://www.multicians.org/fjcc4.html after symlink dereferencing, files themselves are links to an inode. more indirection can go in FUSE userspace-filesystems, a HTTP daemon before it hits fs-paths, or in a databrowser client before GET - or maybe after in 404-fallback-to-cache. cache-populating browser-extensions or offthepkgmanager proxies, replicating a POSIX-fs layer holding RDF files with rsync/scp/ceph/btsync.. in a "dumb" server scenario, there's all this patchability, and most of it optimized by kernel-developers - and bleeding edge distributed-filesystem tools in the staging / git trees - so interested in hearing how you get creative much of what maps HTTP requests to a representation of RDF sourced from files on POSIX is simple URI-arithmetic using stdlib URI-class functions, and within the hierpart of the URI using path-utilities basename, dirname.. most webservers i'll use inadvertently act as a cache/proxy because there's no check to the Host: claims in a request, they just try to find that resource 1. http://man.whats-your.name/mv
Received on Friday, 29 August 2014 22:10:40 UTC