jc / digestif Goto Github PK

Error logs are not checked enough. Sending email will help.

Errors in the log files suggest that Flickr can return an unexpected response. The site breaks because we don't have failsafes in place. I've seen this happen when trying to retrieve the name of the photostream.

At the very least we should not return server 500.

• Log the output of API to be able to program against the output
• Return a failover string (flickr id? instagram id) when get an unexpected response from the API

Idea: if we notice digests are not being opened (mandrill api call for opened status, or url never visited) we should send an email reminding them to check the spam folder or to unsubscribe if they are no longer interested.

Limit the number of times this email is sent.

All private keys and data are hardcoded. In order to push the code into a public repository this data should be placed in a configuration file which is not committed.

Configuration for the following, also consider ramifications of regenerating these keys:

• Flickr consumer key and secret
• Instagram consumer key and secret
• Mandrill API key
• Sendgrid API username and password
• Flask secret
• hashids secret
• MySQL database URI (includes username and password)

Some subscribers have been added to the rejection blacklist. The digest is never marked as delivered and is sent again every day.

Possibilities:

1. add a rejection count and once the count exceeds k mark do not attempt to resend
2. on rejection just discard the digest and update the subscription sent dates
3. on rejection remove the user

...

See http://code.flickr.net/2014/04/30/flickr-api-going-ssl-only-on-june-27th-2014/