Closed realstraw closed 4 years ago
@nburke-salesforce there are actually many issues I discovered when I started to look closer to the implementation https://github.com/krux/starport/pull/57 and start testing it. To name a few:
retrieve
is called, all the pipeline it tries to schedule would be double scheduled, which can be very problematic.There are other issues I discovered too making this very hard to maintain. My choices are either roll back or move to a better implementation (which we were originally planned to use, but for various reasons did not go for while adding support for alternative (e.g. SQS) dispatchers.
there is also additional benefit of making new type of (remote) dispatchers much safer, and even enables the ability to have mixed types too (the previous implementation would be either this or that at the starport level).
@realstraw understood! Thank you for the additional context
@realstraw it would also help to understand why the change is being made -- maybe a note in the description, at least, since we're not addressing a known issue, and high-throughput message handling isn't a concern. Why now?