I see at least two problems. The R package quantstrat is poorly documented. And one must have dividends adjusted data. Otherwise the test results will be irrelevant.
It depends, what alternatives do you have? Quantstrat is useful in certain situations, and the authors did their best to give R a useful backtesting package. You are right, the learning curve is a little bit steep since and it's implementation would perhaps be a little bit less awkward in a typical object oriented design. I have switched to Backtrader in Python which is equally as frustrating at times, although better documented. Regardless, any backtesting library you decide to learn will probably have a learning curve, so it depends what you want to use it for. If your goal is to backtest generally simple trading strategies on a single stock, it will do the job quite well. Adjusted data is reletively easy to come by. I recommend the alphavantage api which is also supported in the quantmod library by setting