Is there any performance difference between `"${va...

# announcements

[mathew murphy]

Is there any performance difference between "${var1}${var2}" and var1 + var2 for string construction?

[Ruckus]

The second only works for strings, whereas the first works with any value.
(Also, you can just use "$var1$var2")

[mathew murphy]

Right, I was just wondering which I should prefer when my values are known to be non-null strings.

[mathew murphy]

In Ruby the template substitution is faster than string append.

[Ruckus]

Whichever you like more. Any difference will be negligible. My uneducated guess is they get compiled/optimized to roughly the same thing. Unless you are doing a metric fetch ton of those operations in a hot loop, I can't imagine the difference would contribute to your actual runtime.

[Casey Brooks]

I believe the first is compiled to a StringBuilder under-the-hood. I’m not sure about the second, if it goes to a StringBuilder or does normal String concatenation. But it should either be the same performance-wise or slightly faster

[mathew murphy]

Sounds like templates are probably the way to go then.

[Marko Mitic]

I guess that best thing to do is to open kotlin bytecode, maybe decompile that to Java and see for yourself

[Ruckus]

Don't decompile to Java. The decompiler has some magic code to try to reinterpret string concatenation into friendlier looking code. It's not representative of what it's actually doing. You'll have to read the bytecode directly.

[Mike]

I don't know when it was added. Javac will turn + operations into StringBuilder in 8+ at least. Not sure if it does it in 6 or 7. So could depend on which version of JDK, and could depend on whether Kotlin compiler is turning it into +, or turning it into StringBuilder?

[Cody Engel]

in terms of which to prefer i would prefer string templates, it’s a feature of the language which makes life easier than having to do string concatenation in another way

[Cody Engel]

i can’t imagine the performance difference (if there are any) would be significant enough to do things the old way

[SiebelsTim]

If I remember correctly, the resulting byte code is the same.

[louis993546]

yup, it's the same (with StringBuilder)