On Thu, May 5, 2016 at 2:39 PM, Carl Meyer <c...@oddbird.net> wrote: > Hi Andrew, > > On 05/05/2016 02:19 PM, Andrew Godwin wrote: > > I will put my hand up and say that this sidestepped the DEP process, and > > that's entirely my fault. It was not my intention; I've been working on > > this for over two years, and only last year did I go public with my > > semi-final design and start asking for feedback; I should probably have > > taken it into a DEP then, but failed to. > > This isn't a past-tense question; it's not too late to write a DEP, and > I personally think that a DEP should be written and approved by the > technical board before the channels patch is merged. I actually assumed > that one was still on its way; perhaps I missed some communication at > some point that said it wouldn't be. >
To be honest, I had entirely forgotten the DEP process existed until this thread started up; I'm not sure what to blame this on, but as a member of the tech board I haven't got an email about approving a DEP since last October, so it's been a while. I think my own experience merging migrations is to blame, which went very like how this is currently going and so I probably gravitated towards it. > > I'm sensitive to the fact that you've already put lots of work into this > and time is short if you want to get it into 1.10. On the other hand, > this is precisely why the DEP process exists: to ensure that significant > changes to Django are carefully considered, in public, in a way that > allows those without time to dig into all the details to absorb and > consider the salient high-level points. I think that is precisely what > the channels work needs (in community/process terms), and I think we'd > be very poorly advised to push forward on merging it without an approved > DEP. > > I don't think a channels DEP would need to delve into the details of > precisely which channel backends are currently available, etc; it would > mostly be about justifying the high-level design (and comparing it to > rejected alternatives for solving the same problems). It would focus on > the changes to Django itself, rather than implementation choices of the > initially-available external components. It could probably copy > liberally from (or just be short and heavily reference) the ASGI spec > and/or Channels docs; that's not a problem. > > I'm excited about the potential of channels and ASGI, but I'm also > suspicious of arguments that it is urgent to merge into 1.10 at all > costs. I'm not necessarily opposed to that, if it's ready on time and > the community discussion around a DEP seems to have reached a > satisfactory conclusion. (I also think that the important thing is to > make sure the changes to Django itself aren't painting us into any > corners: as long as ASGI is optional in Django, the external support > components don't need to be fully mature yet; especially if we can do an > import dance to make them optional dependencies, which I think is > preferable.) > > But I also think it would be far better to wait than to rush it in in > the face of reasonable unresolved concerns from the community, and > without an approved DEP. The argument has been that merging it "sends > the right signal to the community," but I have some concern that rushing > the merge could send negative signals about process consistency and > fairness that could easily outweigh any positive signals about Django > "having an async solution." > Part of me does not want to aggravate my RSI by having to write and rush through a DEP in the next 10 days, but I can't deny that you are likely correct that it sends the right signal given that we have the process in place. That said, a couple of decently-sized features (full text search, password validators) have landed recently without one, so I can't entirely feel justified dropping this from 1.10 given that it is fully written, has extensive documentation, a mostly-complete test suite and several fully-worked examples - far more context than a DEP would ever provide. It would feel like a bit of a kick in the teeth, to be honest. Andrew -- You received this message because you are subscribed to the Google Groups "Django developers (Contributions to Django itself)" group. To unsubscribe from this group and stop receiving emails from it, send an email to django-developers+unsubscr...@googlegroups.com. To post to this group, send email to django-developers@googlegroups.com. Visit this group at https://groups.google.com/group/django-developers. To view this discussion on the web visit https://groups.google.com/d/msgid/django-developers/CAFwN1ur1yOczrOppAKw2b8GqnCsK5CjkBcbRMCOLXKwCnXpCuw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
