[openstreetmap/openstreetmap-website] Note: Add body & author (PR #4481)

[Andy Allan]

> a) Should the data migration (to backfill the notes.{body,author_id,author_ip} run in the scope of the migration or can this be a rake task which is running in the background and allows to perform some sanity checks on the data?

I'm slightly torn on this one. My instinct is that the steps should be decoupled - adding columns in a transaction, and then a separate rake task that can be (re)-run as required to do the migration. In particular, if there's some record somewhere that causes an error (or the process times out or whatever) then I'd rather not be rolling everything back.

However, I'm not sure how this will work for other deployments, who might not be so hot on following the issue tracker and keeping up to date with everything. At some point they will update their installation and run the database migrations, but will they know that there's a rake task to run too? 

At some point in the future, after the migration is completed on osm.org we'll want to do a code cleanup, and probably make `body text NOT NULL`. Perhaps that will be the point that they realise there's something that needs to be done, when that cleanup migration fails? On the plus side, anyone without any notes can run both the migrations fine back-to-back since the cleanup migration would work without error on an empty db.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/openstreetmap-website/pull/4481#issuecomment-1919297704
You are receiving this because you are subscribed to this thread.

Message ID: <openstreetmap/openstreetmap-website/pull/4481/c1919297704 at github.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/rails-dev/attachments/20240131/55459efd/attachment-0001.htm>