I evaluated Go last week (2013-03), but there is no meaningful Eclipse IDE support. goclipse doesn't work half as well as CDT, and not a patch on JDT.
I won't move away from eclipse workflows and project management, the context switch is more expensive than any language benefit.
I'll keep checking for comprehensive Go support in Eclipse, but until then it is a showstopper.
More common main page: http://code.google.com/p/liteide/ Downloads: http://code.google.com/p/golangide/downloads/list Code: https://github.com/visualfc/liteide
You're letting your tool own you rather than you owning your tool.
Autocomplete / suggest broken on Windows. There is something about a fix by running an external server. I gave this a go, sending a few hours piecing together forum comments, but couldn't get it to work 100%.
edit: oops, I misread your comment. At first, I thought you found an alternative. I'd be open to working with you to improve the experience. I do not use windows personally.
It is called gocode (https://github.com/nsf/gocode). Pretty much all Go environments require gocode - the directions for Eclipse (which I don't use anymore) are pretty clear too: http://code.google.com/p/goclipse/wiki/GocodeIntegration
At some stage the context switch becomes too high and you give up. As I said "meaningful Eclipse Support", and digging around to get gocode working harms this effort.
Erlang has a great concurrency model, but it's fairly clumsy for writing procedural code of any complexity, which makes it harder to pick up too.
The libraries maybe (and the runtime). The language, as per semantics and syntax, is more or less the same.
>Why not use C? Because C has no sensible packaging system, is unsafe, requires manual memory management, and offers a degree of control that is unnecessary for most tasks.
But is absolutely essential for most tasks C is used. For all the others, why not use a higher level language like Python?
>Why not use Java? Because it is verbose and picky and basically forces you to use an IDE to get anything done, etc etc.
Some people are not "forced": they choose to use an IDE, and wouldn't have it any other way. It is like telling some Smalltalk guy: "Smalltalk forces you to use the smalltalk image editor/environment" -- so what? He is perfectly fine with it. Heck, lots of people even like using dynamic languages with an IDE (and more would, if there was better type inference and assistance).
>If there were nothing wrong with these languages (they're the main ones we use at Google) we wouldn't have developed Go.
Well, the main reasons for Go, as stated in the official pages (the Golang FAQ), are not very convincing or exciting: No header files, faster compilation, parallelism, garbage collection, lighter type usage.
What C programs are you thinking of? Thinking of significant C programs that I have used (bash, apache, gcc, and many more), I can only think of a few that wouldn't be easier to write and maintain were they written in Go, without suffering a meaningful performance penalty. Funnily enough, a lot of the uptake of Go has been from former users of Python, Ruby, and other dynamic languages; they attracted to Go's simplicity, safety, and speed.
But really there's a lot more to Go the faulty dichotomy of "dynamic vs static" (or Python vs C, whatever). This is worh reading: http://talks.golang.org/2012/splash.article
>lots of people even like using dynamic languages with an IDE
Agreed, and I have nothing against IDEs. I have just found that I can't comfortably write Java code by hand. There's way too much boilerplate. In my experience an IDE is essential writing Java code efficiently, and I don't think that's a good thing.
>not very convincing or exciting
Depends on your definition of exciting. Since I've been writing Go code I've been more excited about programming than ever before.
As much as I love high-level programming, many of the most interesting and important problems are ones where a "small" constant factor is the difference between great and mediocre.
On most algorithmic benchmarks Go is 2-3 or more times slower than C/C++.
And parsing something like a large C++ codebase with a GC language like Go would result in tons of memory used (and slowness).
Heck, even Go compiling Go programs gets OOM'ed sometimes in Linux.
For a project a couple of years ago, I needed to use OpenCV and the Go bindings were quite immature. I just switched to C++ (for that project) and never really returned.
But now that I think about it, whenever people that are interested in being introduced to coding (it happens a lot on campus) ask me "what language should I learn?" I've never answered Go. I'm not sure if most people would.
Certainly not C or C++, if you're discounting Go. I'd also discount Java and C#, since they're syntactically quite similar to Go. Certainly not erlang.
So... python? ruby? While I like those languages, I feel that the lack of a compiler is a pretty big hindrance to new programmers. Statically typed languages can catch many errors that produce some truly strange error messages in dynamic languages.
The language, libraries, and ecosystem are miles ahead of where they were back in 2009.
Why not use C? Because C has no sensible packaging system, is unsafe, requires manual memory management, and offers a degree of control that is unnecessary for most tasks. Why not use Java? Because it is verbose and picky and basically forces you to use an IDE to get anything done, etc etc. If there were nothing wrong with these languages (they're the main ones we use at Google) we wouldn't have developed Go.
Use Go more! You'll see why people love it, and why we created it.