7 ms·

It's worth reiterating that Go was created by Google to solve Google's problems. If it also solves your problems, or a worthwhile subset thereof, then that is a

It's worth reiterating that Go was created by Google to solve Google's problems. If it also solves your problems, or a worthwhile subset thereof, then that is a happy serendipity, but it's not a design goal.

usrbinbash 3y ago
Go was created to solve the problem of how to write real world code solving practical problems in the space of backends, systems (not system) programming and microservices, where performance matters, with a language that is easy to learn, the code of which is easy to read and maintain, which is easy to understand for tooling, and which lends itself naturally to projects where many coders of different skill levels and personal preference work on the same thing. I believe that it is very much a problem many entities that aren't Google like to have a good solution for.
- zozbot234 3y ago
  > easy to learn, ... easy to read and maintain, These things are quite subjective. In practice though, Go lacks the kinds of features that would make very large scale codebases (the sorts of codebases where "many coders ... work on the same thing") truly surveyable, understandable and easily maintained. I mean, it's not a complete disaster like Python or JavaScript. It has some features to support modularity and programming "in the large". But other modern languages like Rust or even Swift are quite a bit better on that side of things, and it shows.
  - usrbinbash 3y ago
    > In practice though In practice though, there is a reason why Go is seeing continuous growth in usage, so I kinda doubt that these features of Go are subjective. > But other modern languages like Rust or even Swift are quite a bit better on that side of things, and it shows. In what metrics does that "show" if I may ask? Because it isn't showing in usage numbers across projects.
  - abenga 3y ago
    And yet the largest Go projects seem to be doing fine, dwarfing the largest Rust or Swift projects in the level of activity, number of contributers, and general impact on the ecosystem. And I speak as someone who prefers to write Rust. Perhaps these features you speak of don't matter as much as we think.
  - ramses0 3y ago
    There's an aside to that: Supposedly within google itself, they have super-magical tooling (ie: rename `dev.foo.bar(abc)` to `abc.dev.v2.foo(bar)`), so it's less important to have those "in the language", when "they" can just refactor "everything". Us mere mortals are trying to pick up Thor's hammer at the end of the day.
TheDong 3y ago
Go solves an industry wide unchanging problem. In the words of Rob Pike: "The key point here is our programmers are Googlers, they’re not researchers. They’re typically, fairly young, fresh out of school, probably learned Java, maybe learned C or C++, probably learned Python. They’re not capable of understanding a brilliant language but we want to use them to build good software. So, the language that we give them has to be easy for them to understand and easy to adopt." – Rob Pike 1 The problem Go solves is that google has a bunch of people who can hack out code, but don't really know the theory of computer science, can't understand high level abstractions and complex type systems... So they need a language for "the average programmer", or as pg would say, "the blub programmer". They invented the blub language. Blub programmers are everywhere at google, and they are everywhere in the industry, so its primary design goal is applicable to almost every company in the world.
- zozbot234 3y ago
  I agree that this is a real problem, but newer languages like Swift and Rust are targeting newbie coders while keeping the benefits of advanced type systems. The key is to surface errors in compiler diagnostics, so that the newbie coders know when they're getting it wrong. It's using the compiler as a friendly TA. Golang doesn't really solve the problems of newbies, it just punts the issues to runtime where they will likely end up causing breakage in production.
  - TheDong 3y ago
    > Golang doesn't really solve the problems of newbies, it just punts the issues to runtime where they will likely end up causing breakage in production. The GC in Go punts the issue of memory management to the runtime, but does so in a largely correct way. In that one way, Go is easier than rust in a way which genuinely does not cause any real production issues. However, the most important tradeoff when talking about Go vs Rust on this startup-infested hellsite is not related to the quality of code at all. The goal of a startup is not to produce working production code, but rather to convince VCs of its trajectory. VCs don't care about buggy code. Every company has buggy code, it's expected you have bugs. Normal. Possibly even a good sign. "Move fast and break things". What matters more is the number of warm programmer butts you have in seats. VCs want you to hit your headcount targets, and unfortunately, hiring programmers that will make your codebase a buggy spaghetti mess is much easier than hiring Rust programmers. Said another way, good code might actually be a problem in startup-land since it will make it harder to meet VCs hiring expectations.
iainmerrick 3y ago
No, it was created by a group of distinguished engineers who just wanted to create a language they liked. As I understand it they wanted to improve on C but take a radically different direction from C++, which they disliked. Rather than being explicitly designed for problems Google was facing, it took a fair amount of time to find a niche (initially, log processing).

usrbinbash 3y ago

Go was created to solve the problem of how to write real world code solving practical problems in the space of backends, systems (not system) programming and microservices, where performance matters, with a language that is easy to learn, the code of which is easy to read and maintain, which is easy to understand for tooling, and which lends itself naturally to projects where many coders of different skill levels and personal preference work on the same thing. I believe that it is very much a problem many entities that aren't Google like to have a good solution for.

zozbot234 3y ago
> easy to learn, ... easy to read and maintain, These things are quite subjective. In practice though, Go lacks the kinds of features that would make very large scale codebases (the sorts of codebases where "many coders ... work on the same thing") truly surveyable, understandable and easily maintained. I mean, it's not a complete disaster like Python or JavaScript. It has some features to support modularity and programming "in the large". But other modern languages like Rust or even Swift are quite a bit better on that side of things, and it shows.
- usrbinbash 3y ago
  > In practice though In practice though, there is a reason why Go is seeing continuous growth in usage, so I kinda doubt that these features of Go are subjective. > But other modern languages like Rust or even Swift are quite a bit better on that side of things, and it shows. In what metrics does that "show" if I may ask? Because it isn't showing in usage numbers across projects.
- abenga 3y ago
  And yet the largest Go projects seem to be doing fine, dwarfing the largest Rust or Swift projects in the level of activity, number of contributers, and general impact on the ecosystem. And I speak as someone who prefers to write Rust. Perhaps these features you speak of don't matter as much as we think.
- ramses0 3y ago
  There's an aside to that: Supposedly within google itself, they have super-magical tooling (ie: rename `dev.foo.bar(abc)` to `abc.dev.v2.foo(bar)`), so it's less important to have those "in the language", when "they" can just refactor "everything". Us mere mortals are trying to pick up Thor's hammer at the end of the day.

TheDong 3y ago

Go solves an industry wide unchanging problem. In the words of Rob Pike: "The key point here is our programmers are Googlers, they’re not researchers. They’re typically, fairly young, fresh out of school, probably learned Java, maybe learned C or C++, probably learned Python. They’re not capable of understanding a brilliant language but we want to use them to build good software. So, the language that we give them has to be easy for them to understand and easy to adopt." – Rob Pike 1 The problem Go solves is that google has a bunch of people who can hack out code, but don't really know the theory of computer science, can't understand high level abstractions and complex type systems... So they need a language for "the average programmer", or as pg would say, "the blub programmer". They invented the blub language. Blub programmers are everywhere at google, and they are everywhere in the industry, so its primary design goal is applicable to almost every company in the world.

zozbot234 3y ago
I agree that this is a real problem, but newer languages like Swift and Rust are targeting newbie coders while keeping the benefits of advanced type systems. The key is to surface errors in compiler diagnostics, so that the newbie coders know when they're getting it wrong. It's using the compiler as a friendly TA. Golang doesn't really solve the problems of newbies, it just punts the issues to runtime where they will likely end up causing breakage in production.
- TheDong 3y ago
  > Golang doesn't really solve the problems of newbies, it just punts the issues to runtime where they will likely end up causing breakage in production. The GC in Go punts the issue of memory management to the runtime, but does so in a largely correct way. In that one way, Go is easier than rust in a way which genuinely does not cause any real production issues. However, the most important tradeoff when talking about Go vs Rust on this startup-infested hellsite is not related to the quality of code at all. The goal of a startup is not to produce working production code, but rather to convince VCs of its trajectory. VCs don't care about buggy code. Every company has buggy code, it's expected you have bugs. Normal. Possibly even a good sign. "Move fast and break things". What matters more is the number of warm programmer butts you have in seats. VCs want you to hit your headcount targets, and unfortunately, hiring programmers that will make your codebase a buggy spaghetti mess is much easier than hiring Rust programmers. Said another way, good code might actually be a problem in startup-land since it will make it harder to meet VCs hiring expectations.

iainmerrick 3y ago

No, it was created by a group of distinguished engineers who just wanted to create a language they liked. As I understand it they wanted to improve on C but take a radically different direction from C++, which they disliked. Rather than being explicitly designed for problems Google was facing, it took a fair amount of time to find a niche (initially, log processing).