Just after celebrating its 4th birthday, Go gets a decimal point upgrade. Can it break out of its niche?
Go is Google's replacement for C and as such it tries to be simple and lightweight. The big problem it seeks to solve is balancing simplicity of implementation while supporting modern programming ideas. Most people who encounter it do think that it does a good job but only as far as it goes. There are lots of sophisticated programming features that it doesn't support and which its designers admit to not knowing how to incorporate into the language without making it like other high-level languages.
For this reason even a decimal point upgrade is viewed as a chance to see what has and has not been included in the language.
The first thing to note is that the Go language team is trying to create updates more frequently. It is only 7 months or so since the release of version 1.1.
The biggest change to the language is the introduction of a new three-index slice. In Go a slice is a portion of an array. It is Go's way of providing what look like dynamic arrays, but implemented as a view onto a static array. For example,
var array  int
creates a static array holding ten integers;
creates a slice that maps to array elements array to array, i.e. up to but not including array. Slices can be resliced so modifying the way they map onto the underlying array. The new syntax allows you to specify a maximum capacity of the slice. For example:
fixes the slice at a maximum of five elements, i.e. 2 to 7, even if it is resliced. This can be used to restrict access to the data in the array beyond a given point.
The other major change is that Goroutines are more strongly preemtively scheduled. This fixes something that looked like a bug in that a Goroutine could hog the system if it didn't return and only one user thread was available. However, for it to work you still have to make a function call as the scheduler is only invoked on entry to a function. At the same time the stack size has been increased and the limit on the number of threads has been removed.
In addition the standard library has been improved and added to and now nil pointers are detected and handled.
Overall the changes are small and not going to satisfy the critics who demand additional features such as generics, exceptions, tail call optimization and so on.
Programmers who try Go seem to like its approach and yet they don't tend to become fanatical ambassadors for it as is the case for Ruby, Python, Haskell etc. Perhaps Go just isn't radical enough or perhaps it is the basic uncertainty of whether it will still be around next year or the year after. Being open source doesn't give it a secure future when Google is holding the reins.