LinkBack Support

I think it would be extremely useful if DEVONthink (Pro version, at least) supported LinkBack (see

Is this something that might be possible with DEVONthink 2?

What do other users think?

I voted Yes. And it’s a big Yes :wink:

I curiously asked about its usefulness last June:

… but didn’t get a response then. No time right now to check later posts where it’s been mentioned.

I’ll abstain from voting until I can decide between “yes” or “maybe”. :slight_smile:

LinkBack hasn’t yet taken off, and the odds are that it probably won’t.

The developer community seems to be sitting back watching to see whether LinkBack achieves enough momentum to justify the development effort to join it, or whether it will die out. Several similar efforts, such as Apple’s own Publish and Subscribe have gone the way of the Dodo. Personally, I liked the similar feature in the old Beagle Brothers word processor and in ClarisWorks.

Maybe some information about the future potential of LinkBack (or an alternative) will show up at the WWDC or soon thereafter.

In the LinkBack link I put in the original post, there is a note that says LinkBack support was added to Stone Create in about 2 hours (scroll down to “Stone Releases Create with LinkBack!” from August 4, 2005 for the original text).

This makes it appear that LinkBack support isn’t too difficult to add. One of your (sort of) competitors, Curio has LinkBack support, and this is a definite "Pro’ for it in the Pros & Cons list I’m creating for the note-taking/brainstorming applications that I am currently reviewing. I would LOVE to see DEVONthink Pro have it as well, as I think it would be a significant asset (if only for the integration with OmniGraffle and (soon) OmniOutliner).

I voted “Maybe”, mostly for these reasons Bill mentioned:

It’s not clear what critical mass LinkBack needs to achieve before more developers are motivated to support it. Might be similar what’s happened with Growl although I’m guessing that’s “much” easier to support than LinkBack.

The real trouble here, it seems to me, is that DEVONthink will always be limited by its ability to incorporate the vast range of file formats that its users use to create content. Everybody’s got a favorite word processor, a favorite citation manager, a favorite outliner, a favorite spreadsheet, and a host of other applications central to a day-to-day workflow. Support for some of these formats come up more often than others on this forum as feature requests (OmniOutliner, Mellel, and Bookends seem to be popular), but I’m not sure there’s an easy solution to the kind of interoperability that most users genuinely need. Thus, DEVONthink ends up being primarily oriented around a small handful of ubiquitous formats like RTF and HTML, plus extremely limited spreadsheet and outliner capabilities.

The question, then, becomes how DEVONthink can remain relatively agnostic about all the formats out there, while allowing users to effectively integrate the data they create into DEVONthink’s database. If LinkBack client capabilities really are as easy to implement as the people behind Stone Create seem to suggest, then I don’t see much of a reason for DEVONthink not to bring it in. Even if LinkBack never expands beyond its current support base, the ability to pull in OmniGraffle, LaTeXiT, Chartsmith, MacSQL, etc. would be more than useful enough to justify the effort if (and perhaps this is a really big if) it really is so simple to do. Plus, if OmniOutliner and Mori follow through with their intentions to add server capabilities, LinkBack would be a major coup for DEVONthink on the outlining front.

Perhaps another thing for DEVONthink to look at would be to allow users to assign XSLT sheets to specific file formats, and thus to at least display their data in human-readable ways. I suppose this is a bit trickier than just that, considering that many of the major XML formats (OpenOffice, OmniOutliner, Mellel, etc.) nest their content inside packages of various kinds, but teaching DEVONthink how to fish these out shouldn’t be too hard. I’m sure many users would jump at the chance to write plugins that would allow their preferred formats to be displayed. Certainly, it’d be an easier approach than getting DEVONthink to render natively every kind of document on the market.

There are enough XML-based formats out there that something along the lines of XSLT plugins would be a commendable step forward. Perhaps those more tech savvy than I know of other/better options. In the meantime, so long as DEVONthink can only handle a portion of my content, I can’t help but feel rather more stuck in Finder-land than I’d care to be. Likewise, my support for LinkBack is less about a belief that will actually provide fabulous functionality, and more about hoping that it’s another step in the right direction.

I can’t say how many people I’ve sold on DEVONthink, only to watch them walk away from it when they realized that it couldn’t handle format x, y, or z. As a content manager, DEVONthink takes a serious usability hit from all the content that can’t be included into it.


Well stated.


Thanks for posting some interesting good points, Peter.

I’m curious how you’re defining “handle” and “my content” in that statement, maybe in the context of wanting to use DEVONthink as a partial Finder replacement more than is current possible? I can try rephrasing that if it’s not clear what I’m trying to understand about how your usage.

A good question. I think I understand your meaning, but let me know if my answer is not as you expected.

On the one hand, I do use DEVONthink as a finder replacement. When I first found the program, I wasn’t looking for much more than a centralized interface for my documents: an easy way to browse through my hefty collection of PDFs, RTFs, and so on. Something like iPhoto for my academic life, I suppose.

That said, though DEVONthink accomplishes all of this in a remarkably elegant way, as I used the program more and more I came to realize that its iPhoto-like qualities are a rather trivial (albeit lovely) part of its much richer search-and-classify functionality. This is, I would expect, a rather typical experience.

So, all of this is to say that I didn’t mean to suggest Finder as somehow parallel to what DEVONthink does, but rather to say that to the extent that DEVONthink can’t handle my content (more on that phrase in a second), I end up needing to fallback on my pre-DEVONthink tools, i.e., Finder/nothing.

“My content”:
My content is nothing terribly unusual, and a good deal of it is already covered by DEVONthink. Articles more often than not end up being in PDF or Word, I typically take my reading notes in RTF, I store a lot of webpages in HTML or web archive, etc. These are the formats that aren’t a problem. Then of course there are the Mellel and OmniGraffle files, my brainstorming sessions in OmniOutliner, charts and mathematical equations, a handful of very proprietary XML formats that I use for linguistic annotation, and so on and so on. The point here is that my content consists of a number of formats that are well handled by DEVONthink, and a dozen or so that are not. Of the ones that are not, some of them are probably common enough to justify built-in DEVONthink support, but several are so obscure that I’d probably be the only one to benefit from it.

As I see it, there are three basic tiers of DEVONthink’s handling. The first is indexing, the second is rendering in human-readable form, and the third is editing. Personally, I’m not very concerned about DEVONthink’s ability to edit much beyond its core set of formats. The functionality is already there for what needs to be there, and it seems like a silly idea to bloat DEVONthink out by trying to teach it to do major word processing, to design webpages, or to remove red-eye from photographs. There are great programs for all of these tasks, and there’s no need to reinvent them all. Indexing and rendering, however, is considerably more important to me. DEVONthink is the best program out there for remembering where an idea is discussed, finding connections between documents, and classifying groups of text files. The usefulness of DEVONthink’s semantic search functions grow as its database does, and thus everything that’s not indexed is missing out on the party.

Of course, I could always print my files to PDF and then load the PDFs into the database, but that is inconvenient enough to be a real problem. On an average day I’ll edit several dozen files as I write articles or go through data, and I really can’t justify bouncing back and forth between intermediate PDFs all day long. Plus, the risk of forgetting and then having a mismatch between the database and the ‘live’ file is serious enough in my particular case.

Content is a tricky beast, and the DT designers have had to make some decisions about what to support and what not to bother with. I have no complaints whatsoever about how they’ve laid down their priorities, but so long as there’s no way for users to implement their own support (via XSLT sheet plugins, for example), people with idiosyncratic format needs are pretty much out of luck. In fact, considering that there’s no way to really handle OpenOffice, OmniOutliner, and Mellel (and surely a slew of others that I don’t personally use), “idiosyncratic” doesn’t need to be all that idiosyncratic at all.

Linkback and an XSLT plugin architecture seem like good options to me, not because they’ll cover everything, but because they seem to offer a good “difficulty of implementation” to “content expansion” ratio. The key here is that neither of them actually require DEVONthink to handle new formats at all, nor really even to understand what the new formats are used for. Linkback uses (I think) some sort of static image embedding based on clipboard contents, and an XSLT plugin could translate the files from a word processor or an outliner into XHTML. In both cases, all DEVONthink would actually be doing is just to provide a generic shoehorn that could facilitate a transparent link between a content format and a way of indexing and rendering it. With a relatively small amount of coding, DEVONthink would see an explosion in the diversity of content it can render and display.

Sorry for the massive response to a straight-forward question. Hope this clarifies something.

You definitely understood what I meant and your answer is much more thorough than I might have expected. Thanks!

There are a few followup points I’d like to mention after thinking them through more clearly so for now I’ll just post this brief acknowledgment.

Wow. Peter expressed my needs almost ideally! What an incredibly well-written set of posts!

I’d love to have DEVONthink Pro be able to import (index and display) my OmniGraffle and OmniOutliner documents, and launch OmniGraffle or OmniOutliner to edit them if I double-click the imported notes’ content.

I think indexing would be relatively straightforward, at least for these two formats, as I think they use XML. Displaying may be tricky; I’m not sure if LinkBack makes that easy or not… Perhaps it sends “subscribing” apps a static “snapshot” of the linked document (as Peter noted)? This would be better than nothing, but could perhaps be cumbersome for large and/or complex documents.

The developers at DEVONtechnologies have certainly proven their abilities, and so I hope that they choose to tackle this challenge. :slight_smile:

Any chance LinkBack support would be re-considered (in the affirmative, i mean :slight_smile:) It’s been a few years since the discussion, and the API is still alive (and recently hosted in GitHub by the Omni folks, no less). I for one would love to have embedded Omnigraffle graphics in my DTP text notes (and easier double-click access to LaTeXiT edition – i know it works with services, but still).