Adding auto create to pub/sub topic.

[dhermes]

Fixes#905.

---

This is labelled don't merge because it doesn't adhere to the truly optimal case described in #905 by @jonparrott. I sent the PR just to have more discussion here (now that @tseaver is back in the US).

I'm worried that implementing the method used by node will be painful. I suppose we could do it with something like a custom class that is just bound to the topic and calls made for it

classCustomHttp(object):
def__init__(self, http, object_to_create):
self._http=httpself._object_to_create=object_to_createself._creation_checked=Falsedefrequest(self, method, uri=None, headers=None, body=None):
# Just do the requesttry:
self._http(method, uri=uri, headers=headers, body=body)
except:
# Do some other stuff

---

BUT, no matter how I slice it, the added complexity doesn't seem to be worth it for a measly feature like auto_create. But maybe I am undervaluing it?

[dhermes]

It looks like I also needed to delete the reviewninja webhook (see #948 for original reviewninja post) from the GitHub settings. Lets hope it stops bothering us 😄

[theacodes]

To be honest, in python looking before you leap isn't as overly onerous as it is in node:

if not topic.exists():
topic.create()

So I'm okay with nixing this if the cost/benefit isn't good. My only issue now is that the check for exists is slow, but that's not this library's fault.

[dhermes]

@jonparrott Sorry for the huge delay in reply.

Since it was @waprin who filed, I want to feel out for how he feels about closing #905.

Also I noticed on googleapis/google-cloud-node#696 that you and @tmatsuo are having some reservations of autoCreate and/or think that exists() is a fine substitute for autoCreate.

[waprin]

Yeah in retrospect auto_create creates more confusion than convenience, I'm fine with just closing them all down.

[dhermes]

Thanks guys! Let me know if there is anything we can do to make the sample writing process easier.