Never automatically derive (de)serialization formats - make it explicit. Better yet, use an IDL (and potentially codegen off that)https://twitter.com/shumovichy/status/1062309831288279041 …
ah yeah, I can see how the implicit linkage between your data types and protocols can burn you if you change one thing without thinking about the other
-
-
This has burned me many many many times. My "favorite" one is I had a Spark job that consumed the output of an upstream job, went on vacation, came back, and it broke. Traced the upstream source code, lo and behold someone refactored a data type with automatic derivation.
- End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
. If you're a client/doing your own internal thing, then
at your own risk. The primary concern is innocent refactoring of a data type (e.g. changing field name) opaquely changing the protocol, screwing clients over