On 03/13/2016 05:55 PM, Stephen Finucane wrote:
create mode 100644 patchwork/migrations/0009_add_submission_model.py
create mode 100644
patchwork/migrations/0010_migrate_data_from_submission_to_patch.py
create mode 100644 patchwork/migrations/0011_remove_temp_fields.py
I see how having these split up makes the code more readable, but aren't
they sort of all-or-nothing?
diff --git a/patchwork/migrations/0010_migrate_data_from_submission_to_patch.py
b/patchwork/migrations/0010_migrate_data_from_submission_to_patch.py
+def create_patch_instances(apps, schema_editor):
+ Submission = apps.get_model('patchwork', 'Submission')
+ Patch = apps.get_model('patchwork', 'Patch')
+
+ for submission in Submission.objects.all():
+ # NOTE(sfinucan) We copy every field _except_ tags, which is
+ # autogenerated anyway
+ patch = Patch(submission_ptr=submission,
+ diff2=submission.diff,
+ commit_ref2=submission.commit_ref,
+ pull_url2=submission.pull_url,
+ delegate2=submission.delegate,
+ state2=submission.state,
+ archived2=submission.archived,
+ hash2=submission.hash)
+ patch.__dict__.update(submission.__dict__)
+ patch.save()
This logic is very time-consuming. It took 20 minutes on my instance. I
just hacked this together in pure SQL to run in about 8 seconds (not
really tested):
migrations.RunSQL(
['''INSERT INTO patchwork_patch
(submission_ptr_id, diff2, commit_ref2, pull_url2,
delegate2_id, state2_id, archived2, hash2)
SELECT id, diff, commit_ref, pull_url, delegate_id,
state_id, archived, hash
FROM patchwork_submission
'''],
['TODO THE reverse migration i haven't figured out yet']
),
_______________________________________________
Patchwork mailing list
[email protected]
https://lists.ozlabs.org/listinfo/patchwork
