On Wed, 24 May 2017 21:36:56 +0200 Raphael Hertzog <hertzog@debian.org> wrote: > On Wed, 24 May 2017, Rhonda D'Vine wrote: > > Also it shines a completely different light on the situation: > > This is different than what Raphael explained. It is not so much > > anymore about avoiding to have to adapt some tracker code because > > the 1.10 version isn't code-compatible (which IMNSHO is a bad > > reasoning for not following the rules), this is much much bigger > > than what came through so far. And > > I was not aware of Neil's problems before today. I'm far from > convinced that the problems he ran into with Lava are generally > applicable to all Django applications. Any sufficiently complex django application will suffer this problem. What is needed is a historical list of migrations and a reasonably complex ORM. It can be triggered without having any user data - as piuparts shows. However, it becomes serious when it causes the installation to fail and blocks access to the user data until the migration problems can be resolved. Downgrading usually does not solve it, a backup would be required (and is strongly recommended in our docs before all upgrades). In other words, it is a real life issue which test applications and unit tests will fail to reproduce. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
Attachment:
pgpG2OxoGCFft.pgp
Description: OpenPGP digital signature