Truly silent option

[lindeloev]

I am creating a notebook including brms examples. brm breaks up code blocks even when silent = TRUE, refresh=0 and the code block suppresses everything ({r, cache=TRUE, results='hide', messages=FALSE, warnings=FALSE, errors=FALSE}). Screenshot:

While a relatively minor problem, this makes it a bit harder to read and copy-paste code. This issue was raised in #166, but maybe the notebook case is more clear? It was also discussed in stan and Rstan, but from what I can read, they fixed it: stan-dev/rstan#49.

[paul-buerkner]

The output you see is send through the message stream so anything the suppresses messages, will suppress this as well. I think you mispelled the arguments of the code block header. It should be message=FALSE and warning=FALSE (no plural).

[lindeloev]

You are right, I misread the knitr documentation. Thank you!

[DanyangDai]

Hi Paul

I was using brm for zero inflated beta in my rmd. With message = FALSE and warning = FALSE, it was still printing a long list of chain fittings. I found the solution from Suppress all output from brms in Markdown files.

Using output <- capture.output(fit <- brm(...)) works well for pdf outputs. Thanks to Yizhang.