Subject says it all. I'm looking for a sql (like) api for ufs. I'd like to SELECT by device, inode, size, group, etc. If all the fs stat data was the db schema, I'd like a query engine that makes indexes of the meta data for fast queries, access to the file data, and queries of offline devices, when their index is available.
I've never heard of an engine built specifically on top of a filesystem for interchangeable use via db api or as regular files. Of course I don't mean philosophically, which we already have, I mean something tight that uses modern db optimizers and enables dual access. A key features would be: - "exactly" the same data/time available via either fs or db queries - Race conditions are avoided through standard practice (atomic process) - Leave the task of journaling etc to the OS - Continuous and simultaneous use of either access mode. Extended features might include: - additional meta data like mime, "tags" and "notes" - additional device descriptors, such as hostname and protocol - file path info (I don't think this is part of the stat schema) - Sure this could be executed in a nosql like manner. I think the extended features could easily be implemented as filesystem data, eg if a device has a particular name, get the proto/host/path data from a file that is on a local dev, perhaps even the db index itself. So, has anyone heard of something along these lines? I think this is fundamentally a pretty simple task. Is there any obvious limitation or complexity I'm missing? I welcome comments off line or to either of the lists I'm posting to (but please not both). -George _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users