Retryer

[ljwolf]

This adds functionality to wait/retry between web requests.

This is necessary for some map APIs, especially when downloading large/high-detail tilesets.

This makes contextily kind of like a general purpose tile downloader for any API in the tile-url format, and I've found it extremely handy to work with rate-limited scientific APIs.

[ljwolf]

Hmm, strange; works locally, but I removed six.BytesIO for io.BytesIO, a three-only construct.

I think this may have something to do with using six.BytesIO rather than the python3 builtin io.BytesIO. Maybe six doesn't implement context handling.

[darribas]

I've finally had a look at the functionality, this is great. If you add (passing) tests, I'll merge this. Hopefully it can make it to the upcoming rc!

[ljwolf]

Not sure why this closed, but I had to reopen this to get the merge commit to show.

Also, not sure how to mock this functionality in a unittest.

[coveralls]

Coverage decreased (-3.5%) to 90.0% when pulling 508da93 on ljwolf:retryer into 8e5039b on darribas:master.

[darribas]

OK fair enough. I guess it is tested it at least works in any other method that pulls down tiles. One more thing, should we then add requests to the requirements.txt? It currently works so I imagine it's required by other libraries (geopy?) but it'd be good to have the explicit requirement. I'll merge afterwards.

[ljwolf]

I don't think requests is required by geopy, but it's used in a large amount of web-touching code. I also use it in cenpy.

I mainly like it for the raise_for_status() functionality.

I've added it to the requirements.

[darribas]

Cool!