Why use functions like CGRectMake?

2019-03-23 08:16发布

I'm curious why functions like CGRectMake and CGPointMake exist and are widely used. when, instead, you can do:

(CGRect){{x, y}, {width, height}}

surely this is more efficient (though I'm guessing not by much) as there is no function call?

Also you can set the origin and size like:

 (CGRect){origin, size}

and as a mixture:

 (CGRect){origin, {width, height}}

What is the reason for not using this, and preferring the Make functions?

3条回答
戒情不戒烟
2楼-- · 2019-03-23 08:59

I suppose it is the same old difference between:

  1. adding a dependency in you code from the internal definition of a data structure;

  2. using a function that encapsulates that knowledge and that could "mask" any changes in the underlying data structure.

In principle, CGRect might evolve into a full-blown class with its origin, size, etc, accessors, etc... I am not saying this would be sensible or that it is likely, only that the reason why you use a function to create instances of a data structure has to do with resilience of your code to changes.

查看更多
倾城 Initia
3楼-- · 2019-03-23 09:06

C99's Compound literal syntax is a somewhat obscure feature. Some people find the Make functions clearer to read, or they find it more familiar. At this stage in toolchain support, it's merely preference.

I'm curious why functions like CGRectMake and CGPointMake exist…

CGRectMake (available since 10.0) preceded OS X's compiler's support of compound literals. Compound literals were officially completed in GCC 3.1 (May 2002).

查看更多
爷、活的狠高调
4楼-- · 2019-03-23 09:11

It isn't more efficient, actually. The CGRectMake function (and others) are declared as static inline, which means the compiler copy pastes the function code straight into every place it's used:

CG_INLINE CGRect
CGRectMake(CGFloat x, CGFloat y, CGFloat width, CGFloat height)

where

#  define CG_INLINE static inline

You go from

// code
CGRect myRect = CGRectMake(1,2,3,4);
// code

to

// code
CGRect myRect;
myRect.origin.x = 1; myRect.origin.y = 2;
myRect.size.width = 3; myRect.size.height = 4;

which in principle is no different from

CGRect myRect = (CGRect){1,2,3,4};

after compiler optimizations and such.

As said above you add a dependency on CGRect being a struct of 4 numbers aligned a certain way as opposed to using a function that has more guarantee to it.

查看更多
登录 后发表回答