Hacker News Reader

Comment by kevinconaway | Hacker News Reader

> asking for monotonic time

His anecdote about Google/Amazon using leap smears to solve the problem is telling. I suspect that they were unable to see outside their own bubble to think about how others may be impacted.

> We did what we always do when there's a problem without a clear solution: we waited

The original problem described the issue and the potential consequences very well and the problem didn't change between the initial report and when cloudflare hit it. It was only until a Serious Industrial User (to borrow a term from Xavier @ OCaml) got bit in a very public way that they actually began thinking about what a clear solution would look like.

View on HN

> We did what we always do when there's a problem without a clear solution: we waited

And this is exactly why the Go designers don't understand language design, and how this ignorance shines through every single place in their language.

Language design is about compromises. There is never a perfect solution, only one that satisfies certain parameters while compromising others. You, the designer, are here to make the tough choice and steer the language in a direction that satisfies your users.

Besides, characterizing their position as "We waited" is very disingenuous. First of all, this is more stalling than waiting, and second, waiting implies they at least acknowledge the problems, which the Go team famously never does. Read how various issues are answered and then summarily closed with smug and condescending tones.

You are being deliberately negative here. Choosing to forego generics in favor of simplicity (and its impact along several axes) is a postcard example of a compromise. It is a tough choice that many people will be unhappy with, but there are also many Go programmers that are extremely satisfied with that direction.

As for acknowledging, well, they have always been very clear about their position. It makes no sense to spend a decade answering the same question over and over with a long and elaborate response which the person asking has already seen and dismissed. I can understand them becoming condescending after a decade of facing people who act with a deliberately obtuse and holier-than-thou attitude.

It's not like they have been lazy - every release of Go has had a large amount of improvements that matter. Working on generics would have meant sacrificing a (probably large) amount of them.

(for the record, I dearly miss generics too!)

Given Google's orientation towards server-side web applications, it makes sense. On the other hand, the real-time OSs such as QNX have had monotonic clocks available for decades, and they use them for all delay and time interval measurements. (In real-time, you use the monotonic clock for almost everything that matters, and the day clock for logging and display only. You don't want the clock used for delays and time intervals to "smear"; it might have physical effects if it's being used to measure RPM or something and seconds got slightly longer.)

Go is great for server-side web applications. The libraries for that are all there and well debugged. Anything else, not so much.

Well, but that's not waiting, that's stalling.

Regarding the leap second bug, I suspect this is an example of perfect being the enemy of the good.

It appeared to me that the golang devs believed so strongly in the superiority of leap second smearing that waiting for everyone to adopt it was better than compromising their API or the implementation of time.Time.

Reply on news.ycombinator.com