<br><br><div class="gmail_quote">2009/3/23 Brandon S. Allbery KF8NH <span dir="ltr"><<a href="mailto:allbery@ece.cmu.edu">allbery@ece.cmu.edu</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="im">On 2009 Mar 24, at 0:21, Michael Mossey wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
I'm a beginner just learning about type inference (through getting lots of error messages and using :t a lot). I find it fascinating, and it seems a shame to disregard this power by declaring all types. However I realize that it's the more global or critical types that should be declared, and more local areas can be left to inference. Perhaps also the compiler's ability to infer is related to its ability to detect compile-time type errors. In other words, if you program a compiler to be sensitive enough to detect all type errors at compile time, then by necessity you are giving it the smarts to infer types. Is that right?<br>
</blockquote>
<br>
<br></div>
It may have the knowledge but not the mechanism (consider C++).<br><font color="#888888">
</font></blockquote></div><br>I guess now would be a bad time to mention that C++0x has type inference :D Off topic, but couldn't resist. Anyway, back to the original topic (kind of), would you consider it bad advice to omit type annotations in languages like ML or F# as well? I can identify with both arguments, and I just wonder why one group of people would recommend omitting them, and another would recommend supplying them, with the only difference being the language. I know in something like F# you have a fairly advanced IDE, so you can simply hover the mouse over a function name and get its inferred type signature. I know you mentioned the performance penalty associated with a type being "too generic", in the sense that typeclass lookups will constantly have to be done at runtime if something is too generic. Perhaps the fact that these other languages don't support typeclasses and hence don't have the same performance penalty is part of the reason why it's ok (or at the very least, not as bad) to do this in these languages as it is in Haskell?<br>
<br>I'm mostly just trying to understand why it's good in one language, and bad in another. I'm sure there must be a number of tangible advantages in each language, that either don't apply or are mapped to disadvantages in the other. <br>