Collections public dumps #513
Conversation
|
Pinging @paramsingh instead of a review request which doesn't work… |
There was a problem hiding this comment.
The changes look reasonable to me, however, I'm wondering if it's a better idea to keep the collections in the same dump. different data dump files make it difficult for users to get into the project, or consume the data, is there a reason why we did the two dumps?
I didn't find any other way to not dump the private collections, other than create a dump without the collections and a dump with the public collections only (consisting of selected rows of three tables). Do you think I could just concatenate the two sql dump files? I must admit I haven't tried that. |
|
I did end up concatenating the two dump files, and it works like a charm, thanks @paramsingh ! I also realized my duplicated tables didn't have their foreign keys, and had to add them to the collection dump sql script. |
|
|
||
| -- duplicate user_collection table with public collections only | ||
|
|
||
| CREATE table if not exists public_user_collection (LIKE bookbrainz.user_collection INCLUDING ALL); |
There was a problem hiding this comment.
I would prefer to have these tables named starting with tmp_ to make it clear that they are dump tables. The clean-public-collection-dump-tables.sql script is a bit terrifying to read without the knowledge that these are not temp tables.
Create two dumps: one without any user collections and one with public user collections only. We can then import both of those files and bob's your uncle!
Replace the existing single file import with importing both the main and user collection dumps, after extracting them from the tarball. Modified instructions and links accordingly
In a transaction, with proper foreing keys, ready to be dumped.
& revert changes in instructions to that of a single .sql.bz2 file
MonkeyDo
force-pushed
the
collections-public-dumps
branch
from
55e4d8f to
447600b
Compare
Problem
With the introduction of user collections, private collections are currently going to be exported in the dumps.
Solution
This PR aims to create a database dump without collections, and another dump with only the public collections.
For that purpose, we create temporary tables (ie
user_collection->public_user_collection) with the appropriate select statements to ignore private collections its items and collaborators, and dump those three tables to a file before removing them.We also want to rename the table names in the collections dump file once it has been created (for example to rename
bookbrainz.public_user_collection->bookbrainz.user_collection