When OCR’ing a PDF failed with the message “not converted” in DT4 beta, I noticed that DT didn’t have full disk access. I fixed that while DT was running. Now, I can’t run any of my smart rules anymore:
Applying rules was always limited to matching items. Version 4 disables now rules in the menu if the current selection does not match their conditions.
The files are in the “Scans” folder on an SMB volume that is indexed into the global inbox. The smart rule should move them into the database (“import” and de-index, if I’m right), then OCR and move them into the global inbox. However, they are still sitting there:
Checking the word count is probably the reason as DEVONthink 4 tries to index PDFs even without a text layer. Does it work after removing this condition?
Abby OCR is installed, albeit not in the DT3 folder, as that has been removed … Regardless, the file remains in the Scan folder but now it is no longer indexed. So, half of the way.
There’s no DEVONthink folder in the system library, only in ~/Library.
And now I understand why: In the “install additional modules”, everything was selected with a blue check box. Which I took to mean “it’s installed”. Now I looked again and clicked on “Install”.
Well… Now it works, including the smart rule. I understand that error reporting can be tough (for DT, I mean) but perhaps a clearer “Install additional modules” dialog and/or an Alert/Notification popping up if one tries to OCR while there’s no OCR installed?
I know, but that’s a bit shaky, in my experience. Sometimes it appears automagically, sometimes not, sometimes on another desktop. And from a UI perspective, an error message as a separate dialog like “You must install the add-ons available under … to be able to OCR” might be clearer. Just saying, I’m a big guy and can figure these things out sometimes. What threw me off this time is that DT4 uses Vision, which made me think that the OCR was successful. Mixed blessing, I guess.
Smart rules are actually supposed to be faceless and run in the background without the user’s attention. In case of Data > OCR an alert might be better indeed.