use native .reverse if available

[rubensayshi]

No description provided.

[dcousens]

can you safe-buffer too?

[rubensayshi]

idk why but tests fail ... src.reverse() is not reversing it o.O

[rubensayshi]

// wtf.js
var a = new Buffer("00ff", 'hex')
console.log(a, a.reverse(), a)

var b = new Buffer("ff00", 'hex')
console.log(b, b.reverse(), b)

var c = new Buffer("00ff", 'hex')
console.log(c)

var d = c.slice()
console.log(d, d.reverse())

<Buffer ff 00> <Buffer ff 00> <Buffer ff 00>
<Buffer 00 ff> <Buffer 00 ff> <Buffer 00 ff>
<Buffer 00 ff>
<Buffer ff 00> <Buffer ff 00>

wtf is going on here o.O

[rubensayshi]

// wtf2.js
var a = new Buffer("00ff", 'hex')
console.log(a)
console.log(a.reverse())
console.log(a)

var b = new Buffer("ff00", 'hex')
console.log(b)
console.log(b.reverse())
console.log(b)

var c = new Buffer("00ff", 'hex')
console.log(c)

var d = c.slice()
console.log(d)
console.log(d.reverse())

<Buffer 00 ff>
<Buffer ff 00>
<Buffer ff 00>
<Buffer ff 00>
<Buffer 00 ff>
<Buffer 00 ff>
<Buffer 00 ff>
<Buffer 00 ff>
<Buffer ff 00>

[rubensayshi]

oh it reverses in-place and when I console.log them together they all the same xD

[fanatid]

yea, moreover slice refer to original buffer, the real copy will be: Buffer.from

var a = new Buffer("00ff", 'hex')
console.log(a, Buffer.from(a).reverse())
// <Buffer 00 ff> <Buffer ff 00>
console.log(a, a.slice().reverse())
// <Buffer ff 00> <Buffer ff 00>

[rubensayshi]

ugh made a mess out of the codestyle ...

[rubensayshi]

there, with safe-buffer / Buffer.from it passes :D

[dcousens]

slice will be [often] a copy in browser, and a view in node.

reverse is in-place everywhere 👍

[dcousens]

allocUnsafe would be fine too

[dcousens]

@calvinmetcalf this had safe-buffer too ha... woops.
Sorry @rubensayshi

[calvinmetcalf]

haha woops

[calvinmetcalf]

can you try rebasing on top of master? if you have trouble with that I can do it for you

[dcousens]

@rubensayshi when using Buffer.from(buf).reverse(), there was always a 100% performance penalty... (2x slower) I don't think this is worth it.

For the inplace case, it is about 30% faster.

[calvinmetcalf]

so it sounds like we just want this for the inplace one, which nicely wouldn't conflict with the safebuffer stuff