I am getting a cryptic message when reading a .csv file with readr 1.4.0, providing the col_types argument:

Error in read_tokens_(data, tokenizer, col_specs, col_names, locale_, : Invalid input type, expected 'character' actual 'double'

The same file reads correctly:

• Without providing the col_types argument
• In readr 1.3.1 with the exact same call (including the col_types argument)

I cannot share the file as it contains patient data and don't have a better reprex, but I am sharing it here in hope others who encounter the error will find this and learn that they may work around it by installing 1.3.1 and hopefully they will be able to share their files. I understand that it is hard to take action on the issue without a reprex. Sorry if this is not appropriate and thanks for all the hard work on the package.