How do track or deal with multiple executions that can change the results?

[yl4070]

A while ago, I ran into a situation where if I run a certain cell multiple times, the results will be different. I had the feeling that I need to rerun all the cells, to really guarantee that my final results are correct. This kind of defeat part of the purpose of using Pluto over Jupyter. The simplest example of this will be

#cell 1
x = 1
#cell 2
x += 1

The real code can be more subtle to detect, especially when the number of cells increases. My question is, how do we assure the correctness of the results for code like this, beside avoiding it if possible? Thanks.

[fonsp]

Hi! The code you sent does not run in Pluto (try it!), could you give an example that's closer to your real code?

[yl4070]

Oh, sorry for that. I was experimenting something with vectors, but somehow decided to drop vector. Nevertheless, the code like below would work.

#cell 1
x = [1]
#cell 2 
x .+= 1

I think my real code resembles something like:

#! cell 1
x = [1,2]
#! cell 2
s = []
#! cell 3
f = el -> el > 1 ? push!(s, el) : nothing
#! cell 4
f.(x)

Then, every time cell 4 is executed, s will have one more elements.