Since WinFS is being delayed, Longhorn won't have full database capabilities in its filesystem when it ships; however, it will still provide integrated support for fast indexing. An interview with Bill Gates today regarding the latest news (the delay of WinFS and Longhorn's ETA of 2006) mentioned that Longhorn's initial indexing capabilities will derive from a combination of Office's existing fast indexing support and some new search technology work going on in MSN.
While I was savoring Apple's Spotlight technology in next year's Tiger release of Mac OS, I forgot that Windows XP already provides fast indexing through Indexing Services and that Office had a variety of indexing solutions available since Office 95.
Indexing Services was really designed for use with IIS to provide search functionality for web sites, but you can use it to index and search your file system. Indexing Services needs to be turned on by enabling the "Indexing Service" and setting the Startup Type to "Automatic" on the Services tool found under the Adminstrative Tools program group.
Unfortunately, fast indexing is not currently integrated into Explorer's Search function or rest of the shell. You have to go to Administrative Tools, launch the Computer Management Console and Select "Services and Applications/IndexingService/System/Query the Catalog" to get a web-based interface to search your file system. There are very few options available, not even the ability to filter the search directory.
If you have Office on Windows 2000 or XP, you are in luck. The "File Search" command in the File Menu will use Indexing Service if it's enabled, but it's not on by default. You might also have to turn on "Fast Searching;" I am not sure if Office automatically detects the presence of Indexing Services.
I was pleasantly surprise to discover that Indexing Service consumes virtually no CPU time. The first few minutes when Indexing Service is first installed is CPU-intensive, since it needs to index the whole drive, but after that, the only time the service ever kicks in for indexing is when files of certain types are created or modified. Actually, I have never seen the CPU utilization level for the Indexing Service (cisvc.exe) above 0%.
The nonexistent overhead of Indexing Services serves as a sharp contrast to my experience with both WinFS and Find Fast, the original indexing technology used in Office 95. (Later versions of Office now use Indexing Services.) The benefits of fast results that Find Fast offers was negated by the processing time consumed by indexing, while not searching. Similarly, the Windows File System service produces a noticeable degradation in Operating System performance, especially in machines with a low amount of memory (ie, 256MB).
There's probably money to be made for someone to provide shell integration of the fast searching capabilities inside Indexing Services, just as the recently acquired Lookout offered fast searching for Outlook.
Actually, LookOut also can index your hard drive and I've found that it is a great way to quickly find files on my hard drive. It is also fairly efficient since I never notice it actually indexing .. it uses idle time to index and shuts down as soon as you start using the pc.
Posted by: Sanjay | August 28, 2004 at 11:13 PM
BTW Explorer's search functionality does use the indexing service if it has been enabled and has done so for the last 4-5 years since Windows 2000 was released.
Try searching for a piece of text on two machines with hundreds or thousands of documents with one with the indexing service enabled and the other without.
The Explorer search appears to return the results from the indexing service almost immediately but generally continues with a regular brute force search as well. Probably to cover files that haven't been indexed yet and/or files for which there is no indexing IFilter implementation I'd guess.
http://www.microsoft.com/windows2000/en/professional/help/default.asp?url=/windows2000/en/professional/help/SAG_INDEXtopnode.htm
My main complaint is that MS doesn't ship the indexing service in XP with a larger set of IFilter's. They only cover Office doc formats, plain text and html. They don't include (licensed or otherwise) IFilter's for other popular document file formats, e.g. Adobe PDFs etc. which some of the 3rd party indexing services do.
BTW, you can download an Adobe PDF IFilter implementation from Adobe. Just a bit buggy to install sometimes.
Posted by: Sean McLeod | August 29, 2004 at 01:48 PM
My personal favourite at the moment is Copernic. This has the advantage over X1 of being free and over LookOut of not requiring to be hosted within Outlook. It also has a nicer, more polished interface and a very handy desktop searching facility. You can also search the Web (using AllTheWeb) from within the program. I've previously been using LookOut but I'm going to be uninstalling it and moving to Copernic.
Posted by: Stuart Radcliffe | September 03, 2004 at 05:11 AM
It's also worth noting that when searching for "A word or phrase in the file" with XP's search function, unless you tweak the properties of the indexing service, it won't search for that text inside files with unfamiliar extensions. Even though the indexing service isn't running! It used to work properly in Windows 98, not sure about Win2k.
The tweak:
Start>Run...>ciadv.msc (OK)
Right-click in white space>Properties>Tick "Index files with unknown extensions"
Posted by: Sam Prince | September 03, 2004 at 03:13 PM
Sam: Good tip on forcing Indexing Service to index files with unknown extensions. Already knew of that one myself, but the fact that it's non-obvious turns many people off to the tool.
The reason I finally turned Indexing Service off is that it's an either/or proposition. If you use IS, you can't choose to use the regular search to look inside of files on a per-search basis. Every so often, I find myself needing to locate a system file that mentions a particular executable name, or something obscure like that. IS made this impossible.
By using a combination of Copernic Desktop Search and the built-in Windows search tool (w/o IS) I get the best of both worlds. The ability of CDS to index both my Outlook folders and message attachments is especially nice.
Posted by: Adam M. | October 18, 2004 at 08:14 AM
I found some of the same things. I posted my comparison of Copernic, Yahoo! and Google Desktop Searches. Microsoft (without the Toolbar Search) did just fine, but not because it's great, just because nothing is very good yet.
The comparison is at http://www.livejournal.com/users/bloggit/3764.html
Posted by: Tony | March 20, 2005 at 08:44 PM