OSNews

Also quite similar to Beagle in existing Linux, and the Mac OS-X search engine. The difference to WinFS is, that Beagle is not using a relational database. However, inotify, the kernel level thing that makes Beagle possible, could easily be made to use other types of storage engines.

E.g. it would only take a few hundred lines of code to connect it to e.g. Postgresql and tsearch2 to get one of the fastest full text search enabled system in the market. The advantage over Beagle, apart from speed, would be that data could be stored and accessed from a central server.

Creating things like WinFS isn't the rocket science thing that Microsoft want us to believe. Once Microsoft finally decides to make use WinFS, the competitive advantage will already be gone.

I in fact am believing that MS already has lost the advantages (if there were any). Look at beryl for instance. AERO, what ?

Security always has been a problem compared to other engineered OSes.

Performance wise, same story. Stability wise, same story...

Edited 2006-12-31 17:40

You make the desktop search functionality seem like a toy project level issue.

Unfortunately it's not. Even if you were indexing only regular files on the disk, you'd need to extract information from them. More specifically learning ~/docs/thesis.pdf has changed is not enough. You'll also need to extract the text and metadata from that document. You'll need those kinds of filters for all file types (multimedia, office, source code, etc)

And you'll not only scan simple files, but you'll have to understand several program configurations. For example you'll also need to read (i.e. parse) Thunderbird profiles, or read Blam! RSS feed storage intelligently.

Furthermore, people are already working on more exciting things, like extracting "semantic relations" and automatically generating queries for you:
http://www.nat.org/dashboard/

What WinFS tried to achieve was removing directories all together (at least logically), but defining a good metaphor is not easy, thus they failed. Otherwise Beagle level functionality already exists in Vista as desktop search.

Creating things like WinFS isn't the rocket science thing that Microsoft want us to believe. Once Microsoft finally decides to make use WinFS, the competitive advantage will already be gone.

True, and that is the case; the problem they face is this; they *could* have released WinFs, but the problem is; it would be just yet another 'me too' searching technology with no real edge over what is offered in Linux via Beagle or MacOS X via spotlight.

What Microsoft was hoping is they could do 'natural language searching' that is, you can plow in an every day sentence like, "find me all documents older than 24 of december written to Santa Claus' and it will spit out all documents that fit that description.

What they failed to grasp is just how complex something like that can be, as it brings in a level of fuzziness and randomness; unlike Beagle and Spotlight which have fixed statements that people must use, Microsoft wanted any old statement being used by the user, and still bring back the same results.

I think its a matter of Microsoft dreaming too much rather than simply, 'lets get the basic foundation on the operating system *THEN* work on refining it further and offer updates later'

BeOS had nothing like WinFS 10 years ago.

It was, however, very similar to what we already have now in Vista and OSX.

It was, however, very similar to what we already have now in Vista and OSX.

No it's not. Where Beagle/Vista/Spotlight allows you to search inside files' contents (besides metadata and filename), BeFS only allows you to search in metadata and filename. This is a severe limitation on BeFS's end, meaning you can only find a specific file if you know its name or metadata contents.

Edited 2006-12-31 20:18