C Stack blows up generating SQL for XGBoost model

[sdcharle]

I have an XGBoost model with these characteristics:

niter: 1436

nfeatures : 143

While tidypredict seems able to parse the model, tidypredict_sql blows up:

tidypredict_sql(model, dbplyr::simulate_mssql())

Error: C stack usage 7969920 is too close to the limit

I get that it's a biggish model, but blow up the stack big? I'm running this on R Studio Workbench on RHEL 8.

Any help is appreciated, the idea of generating predictions in SQL has appeal.