I’m not sure how Yep stores the info exactly but I just switched to “Leap” which is made by the same company. Unlike Yep, Leap is the same thing for all files, rather than just PDFs and they just released their final 1.0 release this week.

Leap has something similar to your “packages” in the form of two features: “File groups” and bookmarks which are smart groups that save a particular combination of tags and other characteristics. This is looking great so far but I continue to send them comments.

If Leap is keeping metadata about files in its DB, they are perhaps using Mac aliases since I am easily able to move files anywhere and the metadata remains unaffected. However, there is one problem: in the current release renaming the files outside the application seems to at least sometimes cause the loss of tags. I have contacted them about this problem.

In Leap there is also the option of saving all the tags in the “Spotlight Comments” of each file – a special tag attached to all files in the OS which make it possible to use the OS wide Spotlight search (like Google Desktop for Windows) and find these tagged files. These are preserved after both moving location and renaming of files.

The original files are kept under ’42Tags’ folder, this makes it very easy to backup. You can always access the file with regular explore as well.

One advantage over the ‘yep’ solution is the introduction of the ‘package’ notion which you can add
few files (pdf, tif, doc, other) tag them as a package and later you will find them together.
(e.g. ‘My CV’ – cv.doc, recommendation1.jpg, etc..)

You can see a video demonstration at our site http://www.42tags.com/video.htm

PS – I have a question Munnin regarding yep, it is probably keeping links to your files, but what if you move files and folders around, how yep will treat it ?

OCRing the texts work well enough with Acrobat, I did that with all the articles I posted to the journal articles on http://ChinaJapan.org/ so that Google could search the contents of the PDF. As you say the accuracy was hit and miss but still enough to make searching the documents generally useful.

Doing that for all the files is a bit time consuming. Acrobat OCR, even on the fastest Windows machines in the statistics lab at Harvard was slow enough for me not to do it with every downloaded file. Newer online journals usually have an OCR layer already, or even selectable text.