Write a serializer for multipart encoded forms

[humitos]

In #5000 we are manually manipulating the request to send the data using MultipartEncoder directly to avoid an issue that we found when sending large data.

The proposal is a solution, but we think it could be cleaner if we write a Slumber serializer and we use it directly instead of touching the request at that level: https://slumber.readthedocs.io/en/v0.6.0/options.html#serializer

Also, once this serializer is written and tested, we need to double check if #5021 disappears with the usage of a serializer.

[dojutsu-user]

@humitos
I am trying to solve this issue.
But I'm facing some problems
I have made a serializer class but can't make it to work with readthedocs.restapi.client.api.
How to use that?

[humitos]

@dojutsu-user well, I don't exactly know how to do it since I've never done it before, but the documentation linked in the description should guide you.

When instantiating the client, it passes the serializer:

s = serialize.MySerializer()
api = slumber.API("http://example.com/api/v1/", format="pickle", serializer=s)

[dojutsu-user]

@humitos
Actually, slumber is always using the default serializer because the format parameter is not taken into the account. That's why I wasn't able to get this working. I searched a little bit and found that there's an open issue also:
samgiles/slumber#131

So, may be we won't be able to do this with slumber.

[humitos]

Yeah, you are right.

Blocking on samgiles/slumber#148

We could install that particular commit/revision in case we really want to do it like this since the project looks kind of dead unfortunately :(

[agjohnson]

I'm also fine working around slumber for this, though the current solution is a little hacky.