Added types and note to erdos_renyi function

[henrik-wolf]

I added typed-definitions to the docstring of erdos_renyi, as we differentiate the two different definitions of the Erdös-Rényi ensemble by dispatching on the type of the second argument.

As I spent some time with the strange edge case of erdos_renyi(n, 1) != erdos_renyi(n , 1.0) today this feels like a good idea.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (f6f8db6) 97.28% compared to head (11fe08f) 97.28%.
Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #340   +/-   ##
=======================================
  Coverage   97.28%   97.28%           
=======================================
  Files         115      115           
  Lines        6789     6789           
=======================================
  Hits         6605     6605           
  Misses        184      184           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[simonschoelly]

I like these changes.

Eventually we should fix this mess and use something like erdos_renyi(n; probability=0.1) and erdos_renyi(n; num_edges=23).

Approved, but will merge it later, so someone else will have time to comment on this if they have a different opinion.

And don't worry about the failing tests, nightly is currently broken.

[gdalle]

Some notes:

• nightly isn't broken, JET is just more sensitive, so we'll have to fix it ourselves eventually (but not here of course), see Tests failing on nightly #338
• DocStringExtensions.jl might be a good tool in the future to automatically generate typed signatures in the docstrings and avoid this uncertainty

[simonschoelly]

* DocStringExtensions.jl might be a good tool in the future to automatically generate typed signatures in the docstrings and avoid this uncertainty

Afaik it is not always recommended to put the type signature in the doc string. This is anyways a special case, because Integer as subtype of a Real leads to a different behavior - something that I don't think is very desirable.