We'll be getting out the next release of git-source-control (https://github.com/intersystems/git-source-control) this month, which includes support back to 2016.2 via an artifact associated with (some) releases. We haven't produced this for the past few releases but will do so for the next one.
You can follow the project here to be notified about new releases: https://openexchange.intersystems.com/package/Git-for-Shared-Development...
Two major things: the data structures at your disposal and when and how to use them, and the fact that your code is in the database (which enables some super cool things but also makes other things harder). Another way to put this would be both "how to use (much) older/more basic language features and functions well" (as others have alluded to with the command/function references) and "how to get the most out of object-oriented programming and IRIS' unique strengths" (particularly, having object-orientedness and all your code right in the database).
Re: data structures, basically you have $listbuild lists (which are just linked lists) and the swiss army knife of sparse arrays. If you're building or working with a simple list of any sort of datatype, or with delimited strings as input/output, $listbuild lists are super helpful and more natural to work with than strings. If you're doing pretty much anything else, you want an array. For general coding and algorithms, these are simpler to use (once you know the syntax) and faster (at runtime) than the %Library.List*/Array* classes you're likely to gravitate toward. That is not *at all* to say that you should write your code all your code in .MAC routines like it's the 1990s - although, depending on the environment you're learning in, you might see some of that.
The simplest value proposition of IRIS (in my opinion) from a general development standpoint is:
Write a class (that extends %Persistent), compile it, you get a table - then you can naturally work with the same data from an object-oriented or relational perspective, whichever is most natural. Both of these models are just thin wrappers around our super fast under-the-hood data structure. Every database has one of those, but with ours (called "globals"), if you want that extra 5% performance boost you can see and work with it directly (although you shouldn't really need to).
But on top of that, your code is in the database, and that means you can interact with your own code from an object-oriented or relational perspective using classes in the %Dictionary package. This enables all sorts of amazing metaprogramming techniques that end up being more natural in ObjectScript than any other language I've worked with. (Disclaimer: I've spent way more time working in ObjectScript than any other language, so take this with a grain of salt.)
The drawback to your code being in the database is that you're probably used to it being on your filesystem, and that means you may need to think differently about source control than you usually would. If you're working through interoperability editors or with IRIS BI, you're already modifying code in the database directly, and should adopt an "embedded source control" workflow - I'd recommend https://github.com/intersystems/git-source-control for this if you're not sure where to start. If you're not using interoperability or IRIS BI, you can work more or less the way you're used to with VSCode, keeping in mind that what really matters is what's running / has been saved to the IRIS instance (rather than what's on your filesystem).
Have you just joined a team that's using ObjectScript extensively and has background in it, or are you learning on your own?
The answer about mocking is great.
At the TestCoverage level, by default the tool tracks coverage for the current process only. This prevents noise / pollution of stats from other concurrent use of the system. You can override this (see readme at https://github.com/intersystems/TestCoverage - set tPidList to an empty string), but there are sometimes issues with the line-by-line monitor if you do; #14 has a bit more info on this.
Note - question also posted/answered at https://github.com/intersystems/TestCoverage/issues/33