DataFrames/CSV.read: error on different machines

General Usage
1

Dear folks,
I work on one and the same project on two different machines (both Ubuntu, Julia 1.4). The project-related Project.toml and Manifest.toml are in the project folder and I activate them before running my script. The DataFrames and CSV packages are pinned to:

[336ed68f] CSV v0.6.1 ⚲
[a93c6f00] DataFrames v0.20.2 ⚲

Here is an ME:

using DataFrames
using CSV
df = CSV.read( "myfile.csv"; copycols=true, delim=';' )

On one machine it works, and on the other the following error appears

ArgumentError: provide a valid sink argument, like `using DataFrames; CSV.read(source, DataFrame)

Do you have any ideas why? Are there possibly other mismatching dependencies?
Thanks in advance.

2

I think you forgot to add that you want to convert the CSV data into a DataFrame.

using DataFrames
using CSV
input = CSV.read( "myfile.csv"; copycols=true, delim=';' )   
df = DataFrame(input)
3

No, this is not the error. Newer versions of CSV require a sink argument.

OP, read the error message! It tells you how to fix your problem. do CSV.read(DataFrame, file).

4

I guess the question that OP has is why does this differ across two machines with identical Manifests?

CSV.read() without sink argument was deprecated in v0.7.0 (see here) but should have been working on 0.6 - I just went back and checked, for Julia 1.4.2 and CSV 0.6.2 I could successfully call

CSV.read("myfile.csv")

as you expected.

The only thing that comes to mind to explain what you’re seeing is that maybe you had CSV installed in your default environment at version 0.8 or above, and had already loaded the package before you instantiated your code? Actually speaking of which, did you use Pkg.instantiate() to recreate the environment from the Manifest?

3 Likes
5

Thanks, exactly this was the case,… I first worked with CSV 0.8.. I thought that pkg"activate ." and pkg"instantiate" are enough to “overwrite” the current setting. But this does not seem to be the case. Restarting REPL and directly instantiating my project did the job.