Improve database queries to prevent 500 error on /users/watches

[smithellis]

• Use select_related and prefetch_related to avoid extra queries

[akatsoulas]

r+wc

[akatsoulas]

Probably is more efficient to exclude from fetching items that you don't care about.

watches = (
    Watch.objects.filter(
        user=request.user
    )
    .exclude(content_object__isnull=True) 
    .exclude(Q(content_type__model="question") & Q(content_object__is_archived=True))
    .select_related("content_type")
    .prefetch_related("content_object")
    .order_by("content_type", "id")
)

watch_list = paginate(request, watches)

[akatsoulas]

the reason for the Q object is to ensure that it will filter out is_archived attributes only for questions. Otherwise it will potentially raise AttributeError for content_types that don't have that attribute defined in their model

[smithellis]

That's a good catch. Thanks @akatsoulas

[smithellis]

We can't filter on the GenericForeignKey field content_object since it can't automatically generate a reverse. I'm working on a fix.

[smithellis]

I also noticed a potential issue with the POST content making multiple saves vs. a single bulk update, and altered that.