Tinderbox autofetching from DEVONthink indexing the filesystem

I thought I’d settled on a nice workflow to write my thesis while being able to work on text away from Tinderbox.

I have an outline in Tinderbox; I create child notes within that outline, each of which is a distinct prototype for externally-ingested content, exposing the URL attribute. My idea is to create notes in DEVON, which are really files on the filesystem, and copy the link of each to the URL of the child notes in Tinderbox, where they are magically imported courtesy of AutoFetch.

So far, so good.

Two issues have cropped up, and I can’t tell if they’re intended behaviour, or if I have to weave and dodge a little somehow in my process.

  1. When Tinderbox ingests the items from DEVON, the URL field in Tinderbox is erased. I confirmed this in an entirely empty, prototype-free new Tinderbox document. I was not expecting this behaviour, as it’s useful to be able to link directly back to DEVON.

  2. My notes are Markdown files (.md) in DEVON and on the filing system. Tinderbox ingests processed markdown text, as RTF. I’d quite like to keep all those raw asterisks, brackets and parentheses around, though, as my Markdown→rendered text happens at a later stage. Is there a way to stop this? In a way, it makes sense that the two programmes are magically doing this. The obvious solution would be to save .txt in DEVONthink, but then the iaWriter and DEVON magic isn’t so obviously straightforward. Is there a flag in Tinderbox’s attributes to make the AutoFetch just pull raw, unmediated content?

I’d add an example, but it’s ① so basic and ② linked to URLs that are UUIDs on my machine only, making it seem futile.

For clarity are you using DEVONthink watched groups or true AutoFetch? Either way, my notes indicate The $URL needs to be the source of the fetch/sync , i.e. the DEVONthink URL. If you need yo store a Web URL for the note, you can make URL-type user attributes.

So let’s say you’ve added a user attribute of URL type called $SourceURL (don’t forget to dill out a description for why you made the attribute—future self will thank you). Now, to migrate all the old URLs. Use an agent to find all the notes with URLs I the wrong attribute. The agent action is:

$SourceURL = $URL;

Now when you autofetch or use watched folders the $URL is free for its documented use.

No idea on the Markdown problem. Anyone?

Hi Mark,

Thanks for the reply - using direct AutoFetch, group watching would be chaos for me - I’m aiming at very manually targeted imports from DEVON.

As I understand it, and having checked your atbref in advance, the URL is reserved exclusively for the endpoint the content is coming from, ie the devonthink-item schema.

I think the markdown conversion is an issue where it’s doing the right thing at a level of sophistication I didn’t fully appreciate.

Thanks. You are correct, AutoFetch uses the note’s $URL as it’s source. Checking (I’ve got DEVONthink but am inexpert with it) I confirm that DEVONthink pseudo-protocol links (i.e. links into docs in your DEVONthink database) are deleted as a result of the first call to AutoFetch that source. This also not the case with autofetch of a web-based URL.

Whether this is due to the use of pseudo-protocols I don’t know. My hunch is AutoFetch of DEVONthink files is tripping over outcomes intended for other ways of importing/fetching DEVONthink data.

This article, DEVONthink, is my most complete collection. Oddly all the things that might set $URL would appear to set it to the URL you’re already storing so I suspect something is amiss. However,
I suggest dropping an email to @eastgate as I don’t think this is fixable via user settings.


1 Like