Until you run into things like "your partners deploy once every two months" or "the team's deploy is delayed by X due to unforseen changes downstream" or ...
Protobuf is built specifically for Google and Google's way of doing things. Not everyone is built like Google.
Re:gRPC: The Ugly Parts
by lopkeny12ko on Saturday September 07, 2024 @04:58PM
(#41476453)
Well, the core problem is that you shouldn't be deploying as infrequently as every 2 months...you should spend engineering energy to fix that rather than on working around it.
Re:gRPC: The Ugly Parts
by crustycoder on Saturday September 07, 2024 @06:00PM
(#41476721)
Must be nice not to have to live in the real world.
Deploying every two months is not required to live in the real world.
Re:gRPC: The Ugly Parts
by crustycoder on Sunday September 08, 2024 @03:57AM
(#41478907)
Customers who expect interface stability for 12+ months is the real world.
You can deploy a service with a stable interface frequently...
Re:gRPC: The Ugly Parts
by crustycoder on Saturday September 07, 2024 @05:58PM
(#41476709)
Yep, well said.
Re:gRPC: The Ugly Parts
by crustycoder on Saturday September 07, 2024 @05:56PM
(#41476703)
If you think that answer makes even the remotest sense then you really don't know what you are talking about.
I worked at Google; I'm describing to you exactly how the infrastructure worked.
Yes, Google's internal infrastructure is that good. It's easy to get caught up in the "haha cancel products" meme and forget that this company does some serious engineering.
Re:gRPC: The Ugly Parts
by crustycoder on Sunday September 08, 2024 @04:02AM
(#41478926)
What Google does with its internal infrastructure is of limited applicability to the majority of people, where interface stability is the prime directive.
Re:gRPC: The Ugly Parts
by lopkeny12ko on Sunday September 08, 2024 @06:06AM
(#41479423)
Interface stability and "living at head" are not mutually exclusive?
They are not. A good interface rarely ever needs to be changed; the point of live at head is that it's impossible to commit a breaking change as long as there's a test depending on the thing you broke.
You revert the offending commit. That triggers an automatic deploy (or, even more likely, your buggy change gets caught in the canary stage and automatically reverted for you).
The Google philosophy is called "live at head" and it has a bunch of advantages, provided your engineers are disciplined enough to follow through.