Hacker News Reader

Comment by cm3 | Hacker News Reader

Is the 64bit check in CMakeLists correct?

I'm no CMake expert, but this looks odd:

https://github.com/picotorrent/picotorrent/blob/develop/CMak...

    # Get current architecture
    if(CMAKE_SIZEOF_VOID_P EQUAL 8)
        set(PICO_ARCH "x64" )
    else(CMAKE_SIZEOF_VOID_P EQUAL 8)
        set(PICO_ARCH "x86")
    endif(CMAKE_SIZEOF_VOID_P EQUAL 8)

EDIT: this is just CMake's very odd syntax and it's actually a single condition.

View on HN

It should be correct - or at least, it works for what I use it for :)

What is odd about it?

I don't know much about CMake, but to me the two conditions look the same, but maybe this is just how CMake magically works.

Whoa, that took me reading it after each sibling comment here and one additional time to "get it". Why does it require the repetition like that?

Oh, yeah, it seems as it is CMake standard to put the if-expression inside the else-parenthesis as well. It is not an elseif.

Yes, came to the same conclusion, seeing how it's the same "string" in IF, ELSE, and ENDIF. Very, very odd syntax.

I get the syntax now. This is a single condition, but it has to be repeated three times. Odd syntax.

It's very unusual, but at least it tells a reader what it's an else of.

In code-reviews I still occasionally have to complain about the

    if (condition){
        // 100 lines of code
    }else{
        // 20 lines of code
    }

pattern.

As we're dealing with a language that has a proper subroutines and a return statement, this can just be refactored away.

But CMake might not have either of those (I don't know - I have only ever written a single trivial CMakeLists.txt), so it might just force this pattern and then it's helpful to also enforce the else to restate the condition.

That makes sense, though the way this seems to be required to write in CMakeLists isn't maintainable. I can only guess they have diagnostics that would warn if you typo'd that easy to mistype string.

Reply on news.ycombinator.com