Gregory Bataille

Hey @phillbaker, no `run_before` does not solve anything. As the data model is crafted today, you cannot deploy one table before the other. What needs to happen is (for example,...

Ok, so I think the (second) problem is different now. Might not even be a concurrency issue. I have `RefreshToken` in my db that have no `access_token` linked. That tends...

Wow, so found it (I think). Nothing to do with a concurrency issue. In some cases, we are deleting (django delete) access tokens directly... So I think the PR submitted...

well it's not linked to an `update_or_create` no. Django is not at fault here (that I can see), it's just that we need to handle the transactionality cleanly, like in...

@jamesshaw49 since I raised these 2 PR and put them in my production, I have never had the issue again. Note that there was 2 things (that I can recall):...

I'm indeed still on 1.1. Following your comment, I'll wait before moving to 1.2! scratch that, I'm actually still on my own fork. Not on Django 2 yet :( so...

While not strictly the same as #9448 I think they might be solved together. First, like @vivanov-dp said, thanks for adding the `for_each` support for modules. I had been expecting...

I'm guessing it's because you don't have a `.githudrc` in you `root` user `$HOME`. that would be the home used by the system when running with `sudo` wouldn't it?

oh, ok. had never come across `-E` I think I see. So if I summarize - `githudd` is not a workable solution in your environment. That however has been fixed...

@voidus I have release 3.3.0 on hackage