have a look at the documentation. String.codepoints/1 differs from String.graphemes fundamentally, regarding unicode and normalization
hypercore: add the same error on both fields separately. You can't add an error for two fields at once.
serafeim: depends. Unicode has two methods on creating a glyph: either combining multiple characters or using a codepoint for a specific character. E.g. your alpha thingie can be expressed as U+03AC or as U+03B1 + U+0301
in the first case is String.graphemes() == String.codepoints(), in the second case it is not
hm, seems to be an timeout issue. setting (test-wise) :timeout to :infinity solves the problem.
this seems wrong to me… what can I do to solve this problem correctly? I have a long-running transaction (a data import module) which can run several minutes up to an hour. I set timeout: :infinity on the Repo.transaction/2 call, but the database connection seems to break due to timeouts, on random other places
I can solve this by setting timeout: :infinity on the repo config, but that's seems wrong to me
I posted on elixirforum about my problem with the ecto timeouts: https://elixirforum.com/t/timeout-errors-with-long-running-transactions/20879
hmm... can I get access to the stacktrace of the error which caused a process to exit in a handle_info?