On Wed, Mar 27, 2019 at 6:13 AM Patrick Georgi pgeorgi@google.com wrote:
Am Di., 19. März 2019 um 21:53 Uhr schrieb Julius Werner jwerner@chromium.org:
I'm not really a fan of auto-formatters because they can just never be as good as a human in all cases.
As I understand Ron's argument, the idea is to accept being "less good" than any single expert person, because it's traded in for more consistency and less friction over everything code formatting since the Machine Is Right[tm].
I repeat Rob Pike's comment: "nobody likes the output of the Go formatter, everyone likes the Go formatter"
meaning that while the output doesn't always meet everyone's standard of perfection, it removes the arguments based on people's different ideas of what is good. And, plus, none of you agree with me or even each other in ALL cases on what "looks good", so at some point these arguments boil down to "because I like it."
I watched this roll out in the Go community in 2011 or so. gofmt was required. There were lots of arguments. In the end, people realized that it was nice to delegate formatting to a robot, because these arguments get exhausting.
Nobody now remembers a time when robots did not format Go code. Nobody wants to go back.
This is true of many projects, in particular those using Rust or Go.
ron