DESIGN.md

What does S3 stand for?

Option 1. S3(b1, p1) vs Option 2. S3(b1, on=[p1])

Pros for Option 1

• Looks nice for when you have to copy from prefix to prefix. e.g. S3(b, p1) >> CopyLambda >> S3(b, p2) >> MoveLambda vs S3(b, on=[p1]) >> CopyLambda(dest_prefix=p2) >> S3(b, on=[p2]) >> MoveLambda

• Works well for multiple prefixes to different services

# Option 1
p1 = S3(b, p1) >> L1
p2 = S3(b, p2) >> SQS1

# Option 2
S3(b, on=[event(p1, LAMBDA), event(p2, SQS)]) >> [L1, SQS1]
 

I have the unnecessary LAMBDA and SQS in my event function

Cons for Option 1

• Does not look nice for when you have multiple prefixes from the same bucket going to one Lambda or multiple prefixes going to different services e.g. A Lambda and a SQS.

 p1 = S3(b, p1) >> L1
 p2 = S3(b, p2) >> L1
 piper = Pipeline(p1, p2)
 #vs
 s3(b, on=[p1, p2]) >> L1

* Two separate S3 objects for the same bucket. Doesn't convey all the notifications from that S3 bucket at a 
glance.
* We have to use `Pipeline` to combine both of them in the same pipeline. But I'd like to keep the Pipeline concept
 only for multiple parents to same the service. One concept, one API.
* If I have to connect a Lambda e.g. `L0` to this pipeline say with an S3 with two prefixes. In `option 1`, I'll
    * have to connect it one of the s3 objects. This does not give a complete picture
    * have to connect `L0` to a Pipeline representing both S3 objects for each prefix? Unnecessary.  

Other Design Alternatives

• Have an S3 for option2 and make a new object S3Objects for option 1