float interface changes

[ghost]

No description provided.

[Araq]

Float64 is now an alias for 'float' because you were right @ventor3000 and it was hard to work with. I am not sure what to do for ARM etc which prefers float32 for performance. I guess people can introduce their own "fastFloat" aliases.

[ventor3000]

This is a good thing. In my taste, int should also always be an int32 (just the way modern c compilers works). This is because I hate code that works diffrently on diffrent platforms. In the unusual case where someone actullay want a type that has diffrent size on diffrent platforms, it's easy to define such a type with a simple when: statement. Also, this way the types cint and cdouble can be removed to slim stuff up.

[ghost]

Many people use int, which implies that they don't care what size it is. Is saving two characters here and there a big deal?

[ghost]

Float64 is now an alias for 'float' because you were right @ventor3000 and it was hard to work with. I am not sure what to do for ARM etc which prefers float32 for performance. I guess people can introduce their own "fastFloat" aliases.

So is one of those types redundant now? Might as well just make float a simple type alias in system.nim rather than having something like fastFloat. Or maybe it could be implemented temporarily if it's hard to change the way float works, and then a tool could deal with the migration.

[ghost]

@ventor3000 I'm sure cint is 32 bits on many platforms, but surely it can vary.

[stevedonovan]

On Fri, Sep 13, 2013 at 1:12 PM, Erik Johansson Andersson <
notifications@github.com> wrote:
@ventor3000 https://github.com/ventor3000 I'm sure cint is 32 bits on
many platforms, but surely it can vary.

Well, sure. int is not a specified size in C, whereas float is.

[Araq]

please make this against devel and I'll merge