Customising raise_for_status

[samuelcolvin]

It would be great if it was possible to customize raise_for_status to:

• add more information to the exception
• log the error/warning (with extra_info) to sentry etc.
• make the response check more specific, eg. require 200
• adapt the error, eg. to make it picklable, see can't pickle CIMultiDictProxy objects multidict#340

Adding all this to aiohttp would obviously be infeasible, but it would be possible to add another kwarg when creating a client session response_check - a coroutine called when raise_for_status=True to check if a response is valid.

This approach would be backwards compatible.

It would also be useful if check_response were also called for request timeouts, but maybe that's more complicated and would not be backwards compatible.

Information which might be useful when debugging http client requests:

• request url
• redirects
• request headers
• request body
• response headers
• response body
• response time

I don't want to add all these to the standard error, but allowing a coroutine to check responses would allow all this to added if I wanted.

[asvetlov]

The feature request sounds reasonable.

Should response_check parameter to be added for every request() / get() / post() method or to ClientSession constructor only?

The real question is: want we add yet another callback into our API?
Inversion of Control is hard to debug.
Are there alternative approaches?

Don't get me wrong: I'm not strongly against the proposal but want to discuss it carefully first.

[samuelcolvin]

I understand.

Technically this is not another callback, just overwriting _raise_for_status with a custom method I think.

[samuelcolvin]

response_check should definitely be added to ClientSession, if I want custom logic on a request I can do it after response as usual.

[asvetlov]

Fixed by #3892