WIP: Add index support to Meta#1
Conversation
…after annotate().
There was a problem hiding this comment.
@MarkusH - Does this validation seem good to you about what we had discussed. The exact behavior we want to check is if index._name is set but I don't think using the private variable _name over here is a good idea. Any thoughts ?
Also for this validation we don't need to explicitly set the name of the index in the from_model method like here.
There was a problem hiding this comment.
I would do the check in AddIndex() I think.
There was a problem hiding this comment.
I agree that this check is required in AddIndex (and hence I have done that in my recent commit related to migrations) but I would like to know your thoughts on keeping a sanity-check here, because in the autodetector there is no use of a ModelState object with an index which doesn't have a name AKAIK (name is used in Index.__eq__). There might be other use cases of ModelState objects other than in autodetector that I am unaware of, in which case we might not want to keep this check.
There was a problem hiding this comment.
you probably want to handle this afer the generate_altered_index_together() method.
5be720b to
43bfacf
Compare
…tions during test database creation" This reverts commit 157d7f1 since it disables migrations all the time, not just during tests.
…ile names Thanks Tim Graham for the review.
Fixed a regression in 625b8e9: improper short-circuiting could lead to a KeyError when threads concurrently call RegexURLResolver._populate().
Forward port of cb78011 from stable/1.10.x.
Workaround a bug with MariaDB and MySQL native client not stripping the `5.5.5-` prefix.
…onse has Cache-Control: no-store.
Thanks Tim Graham for the review.
Thanks Kevin Graham Foster for the report and Tim Graham for the review.
This will fully remove the dependency of the Index class on it's 'model' attribute Thanks timgraham for suggestion
Thanks timgraham for review and MarkusH for advice
Otherwise indexes made using class based indexes get dropped whenever _remake_table is called.
No description provided.