My guess is that part of this was my fault, not Go's. We had some issues with Go being installed in different locations on the CS department machines than on the campus-wide compute infrastructure machines. Within the projects, we also had a somewhat convoluted directory structure in which to mix official code (e.g., interface definitions) that the students weren't supposed to modify and stub code that they were supposed to change. I hope to simplify it a bit for next year. So I'd take this with a little grain of salt: I expect many fewer compilation problems next year just by having had go1 installed for a year and it becoming stable in our environment.
If they are talking about installation/building on personal machines, not university machines (dgacmu's explanation seems fairly plausible there), they may be calling those things difficult when comparing them to using an installation wizard to install some IDE and clicking on menus to build.
My assessment of hassle between the two would be the opposite, but I suspect it varies greatly.
> they may be calling those things difficult when comparing them to using an installation wizard
The Windows MSI package and the OS X .dmg have wizards.
Ah, fair enough.
There's always the installer at: http://www.golang-book.com/. Sets up the path for you and installs a text editor.
> Also setting up a Go environment on a new machine is a nightmare.
How so? There are binaries for all major platforms: http://golang.org/doc/install
All you have to do is extract the archive and adjust the PATH.
> Compilation in general is a huge, unintuitive mess.
Compiling a Go project is dead simple: it's just "go build". How is this a mess?