I’m wondering if anyone has integrated DEVONThink with Git and GitHub similar to the way you can integrate a Scrivener project in Git.
I have a single database that consists of SQLite queries that I run on my LambTracker Program. I’d rather not move them all out into separate files or into something else as it’s really convenient to have them in a DT database that I move from machine to machine and sync across machines and handhelds.
Since the DT database is a package and contains the files (I do not do any indexing in this database) I’m wondering i I can include it as a repository in Git and then commit changes to GitHub regularly as an easier way to share the code with someone else. Plus I really do need the revision tracking on the queries that Git can provide.
Before I jump of into the deep end of trying to configure the .gitignore file appropriately and get everything working I thought I’d ask if anyone here has a stub repository on GitHub already that does this.
This might be doable with lots of caution (similar to storing a database in a cloud folder) but is definitely not recommended as it might corrupt the metadata database (including the search index and the group hierarchy).
Can you explain how it would corrupt the metadata? I’m thinking of it more as a repository on GitHub to which I am the only one that can add stuff, not like a typical code repository that is open source. The benefit to me would be continuing to document queries as I create them in a tool I am using rather than trying to change my whole workflow.
It’s not a question of multiple users, it’s a question of how Git and GitHub handle packages – which is worth reasearching in the user groups there. I suggest creating a test database that has similar characteristics to your normal database (e.g., use of tags, labels, smart groups, replication, etc.) and try it with Git for an extended test. Personally, I would test for a month. Make sure there is always a working backup somewhere other than the repository.
WIll do. GitHub is successfully handling other packages like Scrivener very well so I really don’t expect any problems. I was hoping to not have to create the .gitignore file myself but I guess not.
FWIW the database in question does not use tags, smart groups or any replicated items at all. It’s a very simple one type of query per plain text note system. Some of the notes have several versions of the queries but most are a single query per note. The ones with multiple version may get split into individual notes anyway, I haven’t decided yet.