This week, I was mostly working on improving the architecture of the
code, in particular by splitting ModelResource into two classes,
Collection (for querysets) and Entry (for individual models). Other
improvements include appropriate HTTP status codes for POST/PUT
requests, the option to use templates for the response formatting
(TemplateResponder), basic pagination (works both with serialized and
templated output) and load_put_and_files, a function that populates
request.PUT and request.FILES via _load_post_and_files.
Compared to my initial timetable, I am in time. Looking at my last
status report, I notice that error handling isn't perfect yet (needs
to be unified) and the option to limit the model fields exposed by the
API and the inclusion of resource URIs in serialized output are not
The main thing I will do next week is to think about and improve
automated URL generation.
I am still working on the first part of my proposal (resources that
correspond directly to models). If you have any thoughts on what
should or should not be possible with such a REST API or if you have
specific comments on the code, please let me know.