So the results are good with the upcoming MBS methods. For a recap, we're testing parsing json, generating compact json, and generating pretty json. The new MBS code "competes" with Xojo.Data and the M_JSON module, so I'm only going to include those results here for comparison. All the test variables are the same: same hardware, same Xojo version, same JSON file.
Xojo.Data parsed in 227ms on Mac and 1,884ms on Windows. M_JSON parsed in 965ms on Mac and 647ms on Windows. And JSONMBS.Convert parsed in 238ms on Mac and 216ms on Windows.
For compact JSON, Xojo.Data generated in 174ms on Mac and 432ms on Windows. M_JSON generated in 1,164ms on Mac and 913ms on Windows. JSONMBS.Convert generated in 402ms on Mac and 459ms on Windows.
And for pretty JSON, Xojo.Data generated in 1,262ms on Mac and 1,031ms on Windows. M_JSON generated in 1,219ms on Mac and 952ms on Windows. JSONMBS.Convert generated in 412ms on Mac and 474ms on Windows.
So these new methods offer some clear advantages. While slower in some tests, if you need to generate pretty JSON, they have a very significant advantage over all other options. For those already using MBS plugins, they're worth switching to. I'm personally going to hold out for API 2.0 before making any changes to my code though.