Plan to Implement: oauth2client.transport module

[dhermes]

FYI @nathanielmanistaatgoogle @jonparrott @waprin I am going to factor out all of the usage of httplib2 into a standalone module so we can have a clean break when the time comes to use a better maintained library.

I plan to start soon to please weigh in ASAP with any comments or concerns or requests for API surface area.

Also see #128 for a more "detailed" (but old) plan.

[nathanielmanistaatgoogle]

No concerns. I think I've heard rumblings about this for a while.

[txomon]

Does this mean we will be able to use our own http clients? I am mostly interested on using aiohttp to do async requests.

This sounds like having the functionality of the oauth2 steps separated from the implementation of the transport, but I don't know if it supports the use case I just mentioned.

[theacodes]

@txomon I'm not sure about async transports, but this will certainly make something like that a lot more feasible in the future.

[txomon]

Some difficulties I have found until now is that the docs say that
implementing request function is enough but doesn't give a signature nor
clear signature.

Making that a little more flexible is highly appreciated =) Thanks!

On Fri, 29 Jul 2016 18:25 Jon Wayne Parrott, [email protected]
wrote:

@txomon https://github.com/txomon I'm not sure about async transports,
but this will certainly make something like that a lot more feasible in the
future.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#554 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAN7mgAroLoSldVTjlGTKtjD5Py7Y9bgks5qaimGgaJpZM4JQDbU
.

[theacodes]

That's the goal. Ultimately we want to support urllib3/requests as the primary transport and deprecate httplib2. To do that, we more or less have to enable users to bring their own http client. Two birds, one stone.

[dhermes]

@txomon There are a few more details in #128. The goal is to define an interface that a BYO transport implementation needs to comply with and then just support that. We'll probably have to support the BYO stuff side-by-side with the default httplib2 stuff for a generation or two.

[dhermes]

Calling this complete and moving discussion back to #128