"Multiple definitions" error message on begin ... end is misleading

[laggvar]

This error message is quite confusing:

By renaming a in one of the two cells, the error disappears, so there is nothing wrong with Base. But this can be quite difficult to troubleshoot in more complex scenarios.

It gets even stranger when renaming the variable to c Then the error only mentions Base and not the variable with the actual error. This seems to happen when using the names c or e but not a, b , d or f.

Tested in Pluto 0.20.0 / Julia 1.11 and partially in Pluto 0.19.40 / Julia 1.9.3.

[fonsp]

Woww so weird that it depends on which letter you use!! What is that about 🙈

[fonsp]

It looks like the problem is with the . in Base.zero: here is one message where cee shows, and one where ceee does not show:

The randomness is caused by set ordering of the symbol names :)

[fonsp]

I fixed the error message display, it now looks like this (no matter the variable name):

Now it also shows Base.zero instead of just Base

[fonsp]

About the confusing error message, in your example the improvement would be to detect that a is doubly defined, but the Base.zero definitions don't matter.

Our code understands this. But when the error is generated, a different (simpler) algorithm determines which variable names to show in the error:

https://github.com/JuliaPluto/PlutoDependencyExplorer.jl/blob/b998a9c771cc5e23b1bd235dfbeb34304c122e5c/src/Errors.jl#L22

And this one catches the Base.zero by accident.

[fonsp]

I improved the error message in #3061