Backing up Zotero
Hi there --
I am a newbie. I tried to backup zotero using the instructions on the FAQs but I can't seem to find the file pathway. I have Windows XP.
Thanks,
Sarah
I am a newbie. I tried to backup zotero using the instructions on the FAQs but I can't seem to find the file pathway. I have Windows XP.
Thanks,
Sarah
http://www.zotero.org/documentation/zotero_over_a_network?s=backup
After following steps 2 and 3 on that page (going to the Preferences screen of Zotero and selecting the tab 'Advanced'), click the "Show Data Directory"-button. An explorer window will open showing you your Zotero data.
Why a text format? My reasons.
- Copying the database simply copies problems in the database.
- Having a text dump permits global text editing.
- I might be able to build a DB merge hack.
- I could possibly repair a text file by hand. Can't do that with a DB
I could probably make an SQL dump work for me, but I am hoping that someone has something better.
This is essentially what our DB Repair Tool is doing: dumping the entire database to text (which often works even if an index is corrupted) and importing it back in. We set PRAGMA auto_vacuum=1 on the new DB first to keep the DB size down.
To answer your implied question, ideal would be something like the RDF format that a restore utility could use to rebuild the DB, keys & all. Such a format avoids much of the need to figure out the table structure or the internal linkages. For example 1->n/1->1 associations could be formatted and included with the main record. Of course, such a format doesn't avoid versioning problems, but if a version header was included the restore tool could tell the user that it doesn't know how to restore the file.
In my misspent youth I have used similar schemes as an intermediate format for repeatable format conversions. Programs would generate the DB text form from whatever input they have, and the restore utility would write the physical DB.One common use was to convert a DB when the logical data models are very similar but the physical data models are different (e.g. normalization changes, new computable-value columns, or DB vendor changes).