Regressions from 0.0.8

[rach]

Thanks @leplatrem to work on this project, I'm glad to see that pyramid_storage is receiving some love recently. There is few regressions from 0.0.8 as new settings for s3 are required, eg: port
If you compare to the previous code (0.0.8) much less setting were required. https://github.com/danjac/pyramid_storage/blob/master/pyramid_storage/s3.py#L29-L90

pyramid_storage/pyramid_storage/s3.py

Lines 29 to 51 in 9c56733

	def from_settings(cls, settings, prefix):
	return cls(access_key=settings[prefix + 'aws.access_key'],
	secret_key=settings[prefix + 'aws.secret_key'],
	bucket_name=settings[prefix + 'aws.bucket'],
	acl=settings.get(prefix + 'aws.default_acl', 'public-read'),
	base_url=settings.get(prefix + 'base_url', ''),
	extensions=settings.get(prefix + 'extensions', 'default'))
	def __init__(self, access_key, secret_key, bucket_name,
	acl=None, base_url='', extensions='default'):
	self.access_key = access_key
	self.secret_key = secret_key
	self.bucket_name = bucket_name
	self.acl = acl
	self.base_url = base_url
	self.extensions = resolve_extensions(extensions)
	def get_connection(self):
	try:
	from boto.s3.connection import S3Connection
	except ImportError:
	raise RuntimeError("You must have boto installed to use s3")
	return S3Connection(self.access_key, self.secret_key)

I don't think that the regressions are big deal but may be worth to update the doc to specify which s3 settings are now required.

[rach]

A bit unrelated but maybe it will be worthwhile to look to make pyramid_storage a wrapper around pyfilesystem to have less work maintaining s3 and also being able to support other storage.
https://github.com/PyFilesystem/pyfilesystem