Search sets / Excluding groups from searches

How about implementing savable search sets?

For example, you could save the search criteria of “Search all contents, with wildcards, no case sensitivity, in the group ‘Clinical trials’”, and select the settings off of a menu?

A related request: what about excluding Groups or sets of groups from searches?

I put all the Matlab manuals into DEVONthink, but they are so huge and have so many words, that they keep coming up in search results (which is rather irritating). I’d love to exclude either specific documents or groups from searching.

I suppose the forthcoming feature of Boolean operators would allow this, in which case, it might be doubly helpful to be able to save Search set criteria.


Saved searches (already on my unsubmitted request list) and the ability to exclude groups from searches (good idea!) would be extremely useful.

I’ve thought saved searches would be much like smart groups.  The main difference might be that smart groups would remain visible in views while saved searches are invoked explicitly from a menu/list.  With Spotlight in Tiger, it may be that a saved search automatically creates a Smart Folder.   I suppose accessing too many Smart Folders will have performance implications, similar to running too many simultaneous searches, but Smart Folders could cache their contents.

Related to that, I’d like a way to toggle group (and item?) visibility.  Sort of how dot-prefixed file and directory names are invisible to Finder by default.  Or hidden files in Windows (with its clunky interface).  Imagine, for example, if iTunes/iPhoto had the ability to hide selected playlists/albums so they wouldn’t clutter the interface and/or interfere with navigation when they’re not needed.  Seems almost a feature of necessity for efficiently managing larger collections of items in non-hierarchical interfaces.

The DEVONthink interface uses a hierarchical organization structure but I think certain enhancements like saved searches, smart groups, and invisible items would improve its non-hierarchical “usability”, like Wiki linking has already opened that possibility.  Hopefully I’ll eventually let things like auto-classify do the majority of “physical filing” work for me, but I’m still too stubborn about making the choice on my own.  There’s a balance between “I don’t care where this goes” and “I need to decide where this goes” with the combination of searching and organizing information.

Search sets might be part of DT 2.0 but probably they’re not necessary as DT 2.0 will also introduce “Smart Groups” - such groups and search sets are basically identical.

Excellent! I can imagine that Smart Groups will be even better than search sets, because they will always existing, sitting in my database hierarchy. I look forward to seeing them.


I’d still like an “invisible bit” that can be set so groups don’t always have to be visible.  Of course there’d need to be a way to set them visible again. :slight_smile:

Such an option is currently quite unlikely but requests from other users might change this  ;)

We’ll I’ll bite.

I think the ability to hide groups or files would be extremely welcome. I’m sorely wanting the ability to assign labels to groups and/or files–the ability to hide or exclusively view certain labels would be fantastic, as it would allow a user to focus on high-priority items or items related only to a specific project.

It’s easy in this regard to think of DT as a photo cataloging software, but for documents. I use iViewMedia Pro for cataloging photos, and the labels functionality is one I cannot do without.

I can tell you that the ability to add/remove OS X labels on DT file items is present in prerelease alphas of DT Pro 1.9, and I believe this feature will be present in DT 1.9 PE. In addition, DT Pro 1.9 in alpha allows one to sort selected items by label.

Then you will be able to do a search, select the results and sort by label, which will give you much of what you asked for.

What remains missing is the ability to search database content by label.

I’ve never used labels much, because I dislike seeing multicolored files all over the place.

But I can see a lot of logical power in allowing the user to define and apply labels as additional metadata tags on database items. I would prefer the ability to temporarily add labels, use them to aggregate or disaggregate items in searches, then return my database to the "no label" state if I wish. The ability to search by label would be important.


Bill, that all sounds good. I’m really looking forward to DT pro.

I think the search by label feature you describe would probably work similarly to what I suggested in my last post–the ability to hide or exclusively show [items with] certain labels. This way, basic label search functionality won’t necessarily be considered a “search”–it would be more of a “view” (if that makes sense).

While I don’t use labels in the Finder, I would definitely use them in DT. I could, for example, assign all files associated with completed projects a “completed” label and have them disappear from working view (but still be present in the database). I can’t imagine how most people wouldn’t find this sort of thing handy.

That said, I do think that eventually labels should be included in any kind of advanced search, to further refine search criteria. In the example above, it would be nice to be able to exclude items with a "completed" label from a search.

Something else comes to mind. It would be handy if DT used icons rather than text colors to designate replicants and copies. If color comes to signify more than one thing in DT (i.e. labels and item-type), that could get confusing. For labels, I would like the item name background to be highlighted in the same way it’s highlighted in the finder. Despite DT’s great searching capabilities, I often just scan the item list, and this would be the easiest way to distinguish one item from another.

In DTPro alpha21 comments I wrote "Searching by label would be nice, of course" and would definitely make use of the interesting related suggestions discussed here.

Ahh, I was thinking about DT’s different visual representations of item attributes (again) yesterday.

Not sure how Finder-like highlighted backgrounds for labels (both selected and unselected, in all views) would look.  Maybe we need some mockup wizards here? :slight_smile:

Yes, we’re trying to visualize how we might apply metadata tags like labels and state to DT data. I think there’s lots of logical potential there. Hope it can be simple and consistent, using colors or other marker symbols that allow the user to define and mark properties of database items.

For grins, look at what can be accomplished in quantum physics by tossing in colors:

Source: Wikipedia; search for gluon, or for quantum chromodynamics.

Heaven help me, my eyes glaze and I get a headache whenever I try to read about quarks – so I do that as little as possible.;D

But I am looking for tools to help manipulate my database, and labels (and perhaps other tags, as well) show promise.