Migrating as an R Shiny dev

[Mkranj]

Hi all, love the work on Positron and I already see some features that could improve my workflow :)

I'm wondering about several things I'm used to in RStudio and would love to be able to access in Positron:

1. When running a shiny app, I'd love for it to open in my default browser instead of positron viewer, without clicking several times every time to move it myself.
2. The Problems pane lists 50+ problems in an app that works, regarding objects that "haven't been defined" but actually are defined in separate files. I think the main issue that the app is in a subfolder of the main workspace.
3. Can testthat be integrated into the Testing pane of Positron? I've used it with python in VSCode before, but I can't find such functionality for R.

[juliasilge]

Thanks for these questions! 🙌

We have some info on better supporting opening Shiny apps in your default browser in #8450. The short answer is that you can do it today if you are building a Shiny app where you can run it by clicking the "Play" button, but not if you launch the app from the R console. Take a look for the current advice, and give a 👍 if you run your apps from the console.

We've made some big improvements in static analysis (i.e. not crying wolf with PROBLEMS like you describe) lately, and I don't think I observe the problem you describe anymore. Can you share what version of Positron you are using?

We do in fact integrate testthat in the Testing UI. Here is what I see in an R package:

Are you not observing that?

[Mkranj]

Thanks for the responses!
I didn't see the Shiny extension, will check that out.
I'm using Positron 2025.8.0. Maybe the issue with problems, as well as with tests, is that I'm not building my app as an R package. The repo consists of multiple apps that reuse some shared code, each shiny app in their own subfolder.
My tests ARE in project_root/tests/testthat, but the repo is not an R package. When I go to the test pane, I get asked to install an extension and couldn't find one that's R/testthat specific. I can still run my tests in the console with testthat::test_dir().

[Mkranj]

I've added a DESCRIPTION and a blank NAMESPACE file to my root to test things, and now the test show up and work as expected! So that's fine by me!

I've installed the Shiny extension and edited the options as per the linked post, so I got it to run in external browser with one click, thanks! The options to do so from console would be appreciated. Also a "stop" icon like in RStudio for the running shiny app would be nice, to not have to CTRL+C it.

As for the Problems pane, now there's no problems in any of my app.R files (where I call the library() functions), but when I open an individual .R file with specific functions I get "No symbol named..." problems for any functions from external libraries or other .R files.
This is probably because, as mentioned, this is not a specific R package with a R/ folder for scripts.
Is there a way to disable these problem detections regarding R objects? Or would that defeat the purpose of the Problems pane?

[juliasilge]

Glad that making the workspace a little bit more like an R package fixed up many of the problems!

We are tracking adding a stop button in #7747, if you would like to add a 👍 or add additional info.

You can totally disable R diagnostics with the setting positron.r.diagnostics.enable, but I would expect that you may well miss the feature overall. Is this repo publicly available, so that we could take a look at it, to understand why you are seeing PROBLEMS in an unexpected situation? Or can you create a minimal public repo that demonstrates the problem?

[Mkranj]

I've made a public repo with similar structure: https://github.com/Mkranj/example_positron_structure

The main unexpected problems are functions imported from packages, here they occur in shared_functions/shared_helpers.R and app2/app2_functions/plotly.R. I have to open the files first for the problems to be recognized.
The functions I defined myself anywhere in the workspace seem okay, they don't show as unrecognized symbols.

In my work, these imported functions show up as >100 problems so I'd like that not to happen :)
If I run the library() commands in the console, the problems go away! But I'd like not to have to do this every time I start up Positron and go through individual app's library() calls.

[juliasilge]

Thank you so much for sharing this example! Super helpful to see

I added this example to #2321 (comment) where we are tracking further static analysis of workspace files. I'm not sure we'll support this specific kind of static analysis in the near term (building out a workspace-level understanding of what files source() what other files). You have a couple of options for workarounds for the time being:

Namespace the functions in your helper files

Do something like this for your helper functions:

df_add_rownumbers <- function(df) {
  dplyr::mutate(df, Row_numbers = row_number())
}

More like an R package

Lean more into treating your workspace like an R package, and make an R/ folder to replace your shared_functions/ folder. You'll need to add something like this as well, so the R package knows where the functions come from:

## usethis namespace: start
#' @import plotly
#' @import dplyr
## usethis namespace: end
NULL

You can read more about this kind of approach at:

• https://mastering-shiny.org/scaling-packaging.html
• https://mjfrigaard.github.io/shiny-app-pkgs/
• many other spots!