Openmeta and Tags within DTP 2.x


since tagging isn’t yet (fully) implemented, do you plan to include support for openmeta?



AFAIK no one from Dtech has publicly commented about OpenMeta (or is it Open Meta?) yet, at least not when I previously mentioned it.

It would be very nice if DevonThink did support OpenMeta. Maybe even tagging DevonThink stuff using

Seconding this, there is a real need for integration of tag apps and apps like Devonthink…


We will have a look at it.

Great news. It’s about time we’ve had some semblance of user created metadata that is somewhat systemwide.

I’m looking forward to TORM (tag once read many). I gotta think that Apple has to be looking at a way of providing such a feature in 10.7 if not sooner. Spotlight is cool and all but if I can’t add my own metadata its power cannot fully be realized.

Tags not being fully implemented was the one thing that kept me from buying DEVONthink

I’m a 1.x devonthink user and I do not upgrade to 2.x (3.x?) until I see user custom fields for object, formatted fields for sheets, and real query for search documents.


Not sure what this should be…

I remember a old thread about tag and field for object. If I had to scan and catalogate paper documents, and I need to store data about, for example, Date of creation, Creator, Receiver, Place stored, Condition, Library… and after I need to see all documents created in 1970, in good condition, and stored in Central Library, I can not use devonthink. And -imho- it is a venial sin for a good program for store and researce paper scanned documents.


I try to avoid tags and all the work associated with them. For most of the work I do in my reference databases, the only tagging I’ve done is semi-rough (often very rough) classification into databases and groups. I never assign keywords to individual documents as they are added. I simply don’t need to do that in DEVONthink for the kind of research I do. That’s why I use DEVONthink in the first place.

But what you have described is a cataloging system, and tagging becomes essential to do that.

Using my approaches for associating notes, and by creating a template for those notes that includes the information cues, e.g., “Date of creation, Creator, Receiver, Place stored, Condition, Library”. I would simply fill in the information for each scanned document.

Now I’ve created searchable metadata about each document I’ve scanned. With the powerful search features of DEVONthink 2 I could filter through all the scans to find “all documents created in 1970, in good condition, and stored in Central Library”. I can do that without special metadata fields or a tagging application. I would have to do a lot of clerical work, creating and filling in the information on my template notes. For example, I would have to type in the name of the library in each note, and so on. Necessary with this approach, but drudgery.

The tagging system coming in DEVONthink 2 could reduce the typing drudgery, although I would have to create a tag for each library (typing the library name only once), then drag the scanned document into the appropriate tag. I haven’t played with what could be done with date ranges, but I suspect there’s room for some interesting logical tricks to handle that.

I think there’s excitement in having some sort of system-wide tagging system because it goes beyond a single tool or application. Frankly every time I create a new folder It feels like i’m in an Edsel when I should be in a Ferrari. I’m basically appending metadata via hierarchy and location and not in the smartest of ways either. My success is incumbent upon me being an efficient architect of the schema that works for me.

I think of tagging as a supplement to tools like DEVONthink. My mother is an attorney and there is so much disparate data that must be aggregated for each case. I think tagging would be a natural for attorney as they could append Case # to all documents and recall them with a simple search without much trouble assuming that your tagging is relatively painless i.e. you’re tagging emails as they come in and documents as you open/save them.

I definitely see tagging as complementary to a product like DTP and not really a replacement as their are bound to be many more organizational features in DTP that go beyond user created metadata.

Things begin to get interesting !


FWIW: Daring Fireball Linked List: OpenMeta Is a Hack

Gruber does not say he agree with that. The discussion is open, and the interesting thing is that Ironic Software developers are willing to collaborate with other independent developers.

Personally (not a developer) I don’t see where “&” tags are less a hack than real metadata tags. As Ironic Software developper say, it is an open source project, and discussion about its implementation is also open.

Actually I add a record replicate for each document I scan, but it is simple a dirty trick. And record fields in devonthink have not format, so 20/10/2006 is a string for devonthink and not a date.

Can you make an example?

Anyway, I still think that a program that scan and keep database of scanned documents has to have custom metadata for catalog, query and report databases. Devon can not only integrate last os features, have to build his own. Imho.


Write that date as 20061020 and it becomes a sortable by date string. Wildcard query can identify all documents for “2006” or even March, 2006.

I add keywords manually to the text of some documents since there is no tagging system. Problem of course is to keep the keywords somewhat consistent. I have a list of them in a DTP doc which I keep open behind the main window. Then periodically I have to copy the keyword list and put it into Nisus to alphabetize it. So it is a bit of a hassle.

But, if I foresee that a document will be relevant to a topic that is not actually named within the doc, I haven’t enough confidence (or experience, I guess) with DTP’s AI system showing “similar” documents, so I use the keywords to make sure that the word I am likely to use as a search term is found in the doc. Or, if a doc applies to some particular idea I have about a topic, then I am sure to get appropriate keywords in there. In other words the tags/keywords may represent what I think about a certain document, its connexions, reliability, etc.—rather than a purely objective topical similarity.

Needless to say I am eager for a tagging system. In fact I found this thread because I was looking to see whether this feature had a chance of appearing soon.

I do recognize that the DTech people have worked hard to give us this amazing AI system, and tagging seems like almost an insult to the work and the AI. Please don’t take it that way! Auto-classify works great.

And the search ability is wonderful. IN fact it works so well that I could give up folders altogether were it not for
a. habit
b. sometimes I want to see what I have that is, say, for aspect B of Project C----not everything about B, just the stuff I have collected for this particular project. So for that alone it is worth it. But it could also be subsumed under a tagging function.

Perhaps that is the bottom line, that tags/keywords/user-supplied metadata can be like comments in the margin of a book, representing the reader/user’s thoughts over and above the visible words.

@cycheney : Normally, tags are comming, and user defined metadata is coming also but not in v2.0. What it is being discussed, is if DTP should or not adopt or try to support the new OpenMeta framework. This framework creates a way to add, store and modify tags systemwide, using some Spotlight workarounds.

Based on still-active Ironic Software forum discussions OpenMeta has some important issues that deserve further understanding and resolution before I’d expect Dtech to seriously consider using it.