[foaf-dev] [Camlistore]: Shit just got awesome.
kidehen at openlinksw.com
Fri Jun 10 11:22:21 CEST 2011
On 6/10/11 8:17 AM, Dan Brickley wrote:
> Fwding this update from http://camlistore.org/ ... Brad always does
> great stuff so do take a look.
> Check the presentation
> http://camlistore.org/talks/2011-05-07-Camlistore-Sao-Paolo/#1 to
> catch up on basic contexts. It stores blobs of data, where some blobs
> are json claims about others (covering schema and sharing, pgp-signed
> Use cases doc http://camlistore.org/docs/uses cites distributed social
> networking as use case. For anyone who used his Brackup tool
> http://code.google.com/p/brackup/ there is some heritage there too.
> There is some glue to wire up MySQL for indexing, maybe someone could
> take a look at whether SPARQL could fit in?
If he/they wire for ODBC, JDBC, instead of MySQL (which has ODBC and
JDBC interfaces), IndexDB, and HTTP (which can deliver SPARQL via SPARQL
Protocol in basic form or via SPARQL Graph Protocol) it will solve many
Otherwise, quite interesting and operates in ODS zone :-)
> From the website, "What is Camlistore?
> Camlistore is:
> a way to store, sync, share, model and back up content
> a work in progress
> Open Source (Apache licensed)
> an acronym for "Content-Addressable Multi-Layer Indexed Storage",
> hinting that Camlistore is about:
> content-addressable storage
> separate interoperable parts (storage, sync, sharing, modeling), with
> well-defined protocols and roles
> your "home directory for the web"
> pro-JSON (yet aggressively format agnostic)
> pro-OpenPGP (for signing claims)
> pro-paranoia and privacy
> ambitious, but ...
> programming language-agnostic (parts and different implementations in
> Go, Python, Java, Perl, Bash, ... the language doesn't matter.) What
> matters is well-defined, simple HTTP interfaces.
> neither "Cloud" nor "Local". happily both."
> ---------- Forwarded message ----------
> From: Brad Fitzpatrick<brad at danga.com>
> Date: 10 June 2011 07:32
> Subject: Shit just got awesome.
> To: camlistore at googlegroups.com
> Camlistore's at a milestone of sorts.
> Some highlights of late, if you're sane and don't follow the commits:
> * the blob store now does sync, sharding, replication, and conditional
> blob routing. very configurable.
> * file schema now supports representing files in a hash tree, not just
> a hash list, so you can store really huge files efficiently.
> * using that, libraries (used by both clients& UI server) now support
> splitting files on rolling checksum boundaries, as made popular by
> librsync and https://github.com/apenwarr/bup/ (who relicensed his
> rolling checksum code for my Go port)
> * indexer& search are coming along, indexing more things and serving
> more queries, as needed by the evolving UI.
> * the UI now shows you your recently-edited objects on the home page
> and has a huge blue button to let you create& edit permanodes,
> * changing their title, adding/removing tags, ...
> * drag& drop file uploads, including calculating the file digest in
> digest exists, verifying all the chunks are still present if so
> (server-side), else uploading it, with the serving optionally doing
> the smart rolling checksum splitting for you, finally creating the top
> file schema blob and the permanode for the file, then adding the file
> permanode to your parent container's permanode. (e.g. adding a file
> to a directory or a photo to a gallery)
> Coming up next, in no particular order:
> * photo gallery, blog, microblog app& templates.
> * cross-instance/Internet sharing& authentication, WebFinger integration,
> * read/write FUSE filesystem (to complement the existing read-only
> FUSE fs for mounting old backups)
> * stand-alone no-installation binaries for Linux, Mac, Windows,
> FreeBSD, and App Engine.
> Hackers wanted!
> - Brad
> foaf-dev mailing list
> foaf-dev at lists.foaf-project.org
More information about the foaf-dev