aBothe
Support multiline compiler error messages
It looks like the lines are treated independently and the second one is interpreted as a warning:
main.d(62): Error: struct main.Foo cannot deduce function from argument types !()(string), candidates are:
main.d(25): main.Foo(Range)
But err, there's more semantics in there, innit: On the first line, it's main.d:62, but then, it's on line 25. Hmm. I wouldn't even consider this as a complete bug due to this phenomenon then. Do you know about further 'semantics' in dmd's, gdc's or dub's error messages? Or may lead me to their code sources?
Line 25 is the line of the candidate.
struct Foo(Range)
{
}
void foo()
{
auto f = Foo(""); // maybe one day..
}
Sure, so in which other way should it be presented to the user? Just as an exception? Or in some different behaviour?
When writing this I thought it would just be pulled into the actual error message. But of course if the candidates are not right next to the code anymore this makes things worse. Maybe with links in the hover message? Not sure.
Links or underlined stuff are probably not allowed. And those candidate hints aren't therefore actual errors but rather information messages, yeah, this might be useful. So it'll not be shown as warning but as info message (if XS allows that).
Meh, there are only errors and warnings that may be added without hacking into the API too much. So I don't think that it's that unacceptable if these candidates become added to the warnings list -- They're also shown directly beneath the previously uttered error message, so all in all it should still be acceptable.
Yeah sure it's no biggie. One could still file an enhancement request for the IDE to add an info category (VS does as well) for these kinds of things. Candidate lists, clang-like fixit messages, static analyzer results (which might also benefit from links between them.. 
)