(actually it returns a new schema, which the surrounding code understands as “please use this schema instead”)
these are mostly lib specific things though. marshallow kinda forces us to do that way, if we want to use. we could also do the JSON dumping differently, use different APIs or do it by hand. The nice thing about schemas is, that it is very explicit what will be rendered and how and you can even use those to “validate” incoming data...
(not that we are doing that at the moment)
yes
alright. bed time now :)
you good for now?
otherwise write stuff and I’ll answer it first thing tomorrow morning.
thanks!
Night!
yes, thrusday is still scheduled for me.
1) You said above that I should a) add a meta tag to my new object and b) register it with the ObjectField registry. What’s confusing me a bit, though, is that I’m not really using a new object, just the existing LinkSchema copied over from the hacker news site. All I’ve done there is just assign an instance of it to a new variable with the many flag set:
2) I think one thing that might be going wrong here is that I’m basically bypassing that chain we discussed yesterday (i.e.
3) So is the right thing to do here to have my view for the
links = LinkSchema(many=True). Then in the views.py, I’d imported links as links_schema, and used that to process what is returned to the client.2) I think one thing that might be going wrong here is that I’m basically bypassing that chain we discussed yesterday (i.e.
query.pagination() -> PaginationSChema -> MorphingSchema -> (target schema)). Is that right? This is perhaps why the data that’s being sent to the client is off: because it’s just the target schema essentially, and hasn’t been wrapped by the PaginationSchema? Am I on the right track?3) So is the right thing to do here to have my view for the
/links/ route return in a way that resembles what you’ve got in that /latest view? I.e. feed a query = Object.query to objects_paged.dump() imported from beavy.models.object, passing it through the as_page() method?
Looking forward to Thursday!
the basic (later) idea would be to have a bunch of listing views which are global and you only configure your LinkObject to be included in them in a specific way (via the Object-Capabilities)
you’d still very much need a Schema though.
but to answer your question, yes, you don’t use the pagination-system supplied at the moment.
I am not certain that is causing that particular bug, but I recommend staying in that chain, to get all the benefits provided.
Huh. Okay. Perhaps, then, I should think of this as a merge into Beavy and work on the global solution of having listing views. The interaction between the back end and the front here is still a confusing area here where I have a lot to learn and that sounds like a good exercise.
gotta do some committing!
beavyHQ/beavy#37
the persona system?
(like … discourse. no difference between their system-bot and any logged in person – is just awkward.)
though I am sure this will come a with a range of other complications :D
those are tests in between the total unit-tests and the functional tests we had before – running a fake app context
(with a database and all)
Regarding test location: I saw you put those in beavy/tests
What’s our guide here?