PR engine relies on reading from github access engine's sandbox. This means that any change in data store implementation of github access engine will impact PR engine as well. Engines should be encapsulated and not leak implementation details between each other. To solve this we could add a GET /token endpoint to github access engine. One problem with this is that if lazy is on the "internets" then token could be easily leaked (completely different situation from deciding not to encrypt the tokens in the data store). To solve that leakage problem we would need to separate internal and external endpoints (see #44)
PR engine relies on reading from github access engine's sandbox. This means that any change in data store implementation of github access engine will impact PR engine as well. Engines should be encapsulated and not leak implementation details between each other. To solve this we could add a
GET /token
endpoint to github access engine. One problem with this is that if lazy is on the "internets" then token could be easily leaked (completely different situation from deciding not to encrypt the tokens in the data store). To solve that leakage problem we would need to separate internal and external endpoints (see #44)