PPC64 To Be Discontinued In Fedora 29, Favoring PPC64LE For POWER

[mulenmar]

If you want a big-endian machine, go pick up a G4/5 Mac and put Debian 8 on it.

My understanding was that the main reason for preferring big endian, aside from it being used in networking and not requiring bit-shifting to create/open packes, was that the bit and byte orderings were consistent -- not anything inherent to the math involved?

[Weasel]

My understanding was that the main reason for preferring big endian, aside from it being used in networking and not requiring bit-shifting to create/open packes, was that the bit and byte orderings were consistent -- not anything inherent to the math involved?

But that's wrong though. Little Endian is more consistent.

When you order bits, you start with either bit 0 or bit 1 (doesn't matter), and that is the least significant bit. Lower bit value = less significance.

Why should bytes be different? Byte 1 (address) should be less significant than Byte 2 (address+1), which is exactly what Little Endian does. For both bits and bytes.


Ah right I know you'll come and say that the least significant bit is rightmost, but you see... that's just flawed human notation There's no right or left, there's just bit or byte positions (or addresses) and that's how you order stuff around.

Like I said, humans write digits backwards, right-to-left instead of left-to-right like we do with words (might have to do with the fact they're arabic digits). You might not actually write them right-to-left, but your brain processes them that way. For example you have to know how large the value is (how many digits) before you write it to make sure it's aligned properly.

Ever wondered why we right-align digit columns while we left-align text? Blame Big Endian in human notation for this ass-backwards confusing crap.

[mulenmar]

But that's wrong though.

I wasn't sure I remembered http://facstaff.bloomu.edu/bobmon/re...Holy_Wars.html correctly, which is why I phrased it as a question. :P

[coder]

My understanding was that the main reason for preferring big endian, aside from it being used in networking and not requiring bit-shifting to create/open packes, was that the bit and byte orderings were consistent -- not anything inherent to the math involved?

Well, think about this: on LE architectures, when I shift a bit off the top of a byte, the next position is the lsb of the next byte in memory. This is true irrespective of what word size I am using. IMO, it doesn't get much more consistent than that.

It only becomes "hard" when using LE to parse bitfields out of a BE bitstream. It's still not bad, though.

Anyway, I wonder if the tide has even started to turn on the "network byte order" convention.

[coder]

I wasn't sure I remembered http://facstaff.bloomu.edu/bobmon/re...Holy_Wars.html correctly, which is why I phrased it as a question. :P

I always assumed Endian was an allusion to Swift. I think byte order (or bytesex, if you prefer) is a better term. Besides being less descriptive, Endian imputes a false sense of validity to both arguments and paints the issue as entirely arbitrary and unnecessary.