holly shhhh how did I miss that! Would that work with assertions erros in exunit? I cannot try right now but that would be great. Especially in december.
lud
stevensonmt
That makes a ton of sense, but does not convince me that the current default is optimal. Feels like optimizing for the people who would know how to get the result they want while making it confusing for the people who have no clue. From my perspective as someone not doing enterprise development I’d rather optimize for reducing the learning curve and then have a configuration option for the more advanced users/developers.
1
LostKobrakai
You can run into this as soon as you deploy your first phoenix app and have it slightly misconfigured - especially if ssl is somehow involved. This is not some niche issue.
2
stevensonmt
I didn’t mean to imply it was niche, but even your example is a few steps beyond the beginner level. Someone configuring a phoenix app should probably be comfortable with other configuration settings where a neophyte should not be expected to know this unique quirk of the language out of the box.
0
LostKobrakai
They‘re not expected to know the that. There‘s a reason charlists default to a sigil in recent elixir versions - to add signal to charlists being something to inform yourself about. Similar to how you‘d look up what ~w does for example.
My argument is not to pretend this is not a problem, but it is about which default is more problematic. In my experience I‘ve run into charlists far more often in errors surfaced from erlang than I‘ve run into a list, which happens to hold smallish integer values, but wasn‘t text. Erlang errors can already be cryptic enough, printing them as lists of numbers would make them far worse. And it‘s not just experienced people running into erlang errors.
4
stevensonmt
I understand. We have had different experiences, which is why we seem to be valuing the default setting differently. The regularity with which newcomers ask the question along with the apparent fact that his is a unique quirk to Erlang among all programming languages everywhere AFAICT makes it feel like an unnecessary stumbling block to newcomers. You disagree on the “unnecessary” part of that statement. That’s fine. But nothing you have said negates my experience.
1
LostKobrakai
Yeah. I’ve seen my fair share of all of those questions. I however don’t think they’ll go away by changing the default. They’d just change to “why the hell do I get those garbage error messages”.
2
sodapopcan
IEx tools are more geared toward ease of development than learning, which makes sense to me. It’s happened at least once where someone’s confusion was caused because they didn’t realize they weren’t getting the whole struct when using IO.inspect or dbg. Those tools, being the ones you first discover, are optimized for debugging over learning which is what I think the charlist default is probably going for as well (based on @LostKobrakai’s thoughts).
I’m indifferent to the default changing (not that’s it’s even on the table) since I never use charlists and I don’t mind the same question being asked over and over. It’s not a bad intro to the forum and we’d otherwise be missing out on amusing answers like @outlog’s 
0
7stud
Check this out:
Interactive Elixir (1.14.4) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)> [97,98,99]
'abc'
iex(2)> [97,98,99,0]
[97, 98, 99, 0]
iex(3)> 'abc' === [97,98,99]
true
iex(4)> Enum.map('abc', fn n -> n*2 end)
[194, 196, 198]
1