I'm concerned about doing this if both values are tuples and there might be a mismatch:

cannot convert value of '(Int, Float, Double, Float, Int)' to expected argument type '(Int, Float, UInt, Float, Int)'

This might not be at the top-level either; the two types might both be pointers to such tuples. Clang has a complicated Type diffing system to try to see when this is the case, but I don't know how it handles implicit conversions.

That's a fair point! I think we should at least handle easy cases here like - if there is a single conflicting element - mention its type and position in custom error message, otherwise try to see if types could be folded. For the nested tuples we could also probably mention them separately and note what position they were in.

It's also the sort of thing we could at least have a frontend flag to turn off, if it turns out the compiler is hiding information that someone needs.

Sounds good!

Comment by Rajagopalan Gangadharan (JIRA)

@xedin @belkadan This seems to be an interesting task. Could I work on this? if yes could you please give some pointers that might be useful. Thank you!

Comment by Rajagopalan Gangadharan (JIRA)

If I understand this right, we should be adding a flag say truncate-type or something. If its turned on Then similar types to be grouped For eg (Int,Int,Int,Int,Float,Float,Float,Float) could be written as

(Int...,Float..) something like this?