Shiny app is unable to deploy

shiny shinyapps.io
#1

Hello I am unable to deploy a shiny-app with the message below:

Preparing to deploy application...DONE
Uploading bundle for application: 7832248...DONE
Deploying bundle: 7170079 for application: 7832248 ...
Waiting for task: 1296973664
building: Parsing manifest
building: Building image: 8523304
building: Fetching packages
building: Installing packages
building: Installing files
building: Pushing image: 8523304
################################ Begin Task Log ################################
################################# End Task Log #################################
Error: Unhandled Exception: Child Task 1296973665 error: Unhandled Exception: 'utf-8' codec can't decode byte 0x8b in position 39887: invalid start byte
Execution halted

Can anybody help to understand the error and debug?

1 Like
#2

I also get the exact same error on my Shiny app based on Golem using MacOS ventura. It seems to be an erratic behavior. 90% of the times I try to deploy this happens, the other 10% it goes through without any change other than just restarting the session. It is very annoying and time consuming. Google searches suggest to remove the .Rhistory files and check for non-ASII characters, I've done that and sometimes works but it never goes totally away. I hope someone can give a better answer than me.

#3

Thanks for you reply!

What are non-ASII characters?

I will look into removing the .Rhistory files.

#4

ASCII is a character encoding with only 128 code points, the first 32 of which are control characters (e.g. tab, newline). 0x8b in decimal is 139, which is too high (i.e. > 127) if something is trying to decode as ASCII. Until about 20 years ago it (or variations of it in the form of ISO-8859) was the predominant encoding used by computers in western locales.

But your error is about UTF-8, which is an encoding that's designed to be able to handle the whole Unicode character space while having the first 128 code points be backward compatible with ASCII. UTF-8 uses a variable number of bytes per character, but not every possible byte can be the first byte of a character. Clearly 0x8b is one that can't be.

The TL;DR is that somehow you're getting an invalid character in your output. I don't think it's diagnosable based on just the error message.

#5

Thank you for your detailed response!

is there anything else that could help diagnose this issue in your opinion?

#6

What would be helpful would be a reproducible example (aka reprex) of code that triggers the error. I know it can be hard if the bug happens intermittently, but if it's even fairly consistent, then try to make a minimal example off your code that gets the same behavior.