Debian Guts Support For Old MIPS CPUs

[carewolf]

Why? There are inherent advantages to BE (e.g. imaging, networking). Quite frankly I'm not sure why LE is even still used, it should have gone extinct long ago.

LE is superior, I see no benefits to BE. LE allows much easier and safer type-casting (the adress of an int pointer as a char or short pointer is unchanged). That BE is used in networking because it was designed on BE unix architectures and it was just the native endian, not for any other reason. Imaging mostly use LE as it needs to be fast on common hardware.

[carewolf]

No. BE is only used in networking by convention - not because it's actually any better-suited to the task.


LE has the advantage that the numeric value of a byte in memory doesn't change based on the word size you use to read it.

The only real advantage BE has is due to the western convention to write more significant digits to the left. So, when viewing a hex dump, it's easier for silly humans to understand. The better solution is to write better tools. For instance, if you just have your LE hex dump display addresses increasing from right-to-left, that would nullify this advantage of BE.

That is all correct.
And the reason we write BE, is because we took arab numbers from a Arabic that writes right-to-left and kept the number right-to-left instead of inverting them to match our left-to-right writing

Though the random coincedence that we write numbers and text with different alignment makes western invoices and budgets better organized since we can align and flush the text fields to the left, and align and flush the numbers to the right. Languages that writes text and numbers in the same direction needs to know the largest number before being able to write the first number in a column on the other side of the paper, and probably wouldn't have invented the notion of flushing numbers to the other side of the paper as it only makes sense when text and numbers are aligned opposingly.

[tuxd3v]

That is all correct.
And the reason we write BE, is because we took arab numbers from a Arabic that writes right-to-left and kept the number right-to-left instead of inverting them to match our left-to-right writing

That is all Wrong..
Network Format is BigEndian..
You can try to paint it has you prefer to suit your ego, or Religion( ..but that is YET another thing.. )
This subject is like a religious war, never ending..
Each architecture has its pros and cons..

Number in BE are correctly printed, in hexa, as it his a natural notation for humans.
ie:
0x0123456789

I am using the Arabic Numerology, like we use in the Christian Society, from the left to right..
A number in BE is like that!

On Contrast,
Writing a number in LE, is written like in Arabic Alphabetic Cultures... from Right to Left.
taking the same example, , LE is like this:
8967452301

This is the motive why is more efficient to convert from Binary to Decimal in BE, than in LE..
Its also a lot more efficient to know if a number is positive or negative in BE..

It doesn't meant that LE arch's have not adapted their instruction sets to solve this network problem,
They indeed for sure have made something about it..but even so, it waists energy doing conversions for every single packet, you receive/send..

Usually the user don't feel this problem, because its the operating system that do this low level stuff, with functions like:
htons() - "Host to Network Short"
htonl() - "Host to Network Long"
ntohs() - "Network to Host Short"
ntohl() - "Network to Host Long"

This subject its like religion, for me from a natural human form, hexa format is very appreciated, and for that BE seems better..
But this has nothing to do with x86 vx MIPS vs others...since this stuff is inherent to formats, nor archs..
Ofcourse LE format will fetch first number( 0x89) to the lower significant register used, it will be almost the same( not for hexa readability..has humans tough ).

But the network problem is there for LE..

[tuxd3v]

No, you're confused.

Am I?

The process of byteswapping is in theory an inefficiency, but only when sending/receiving across the wire AND a lot of standards specifically denote little endian wire formats these days, since most machines are little endian and have be nigh forever.

You are talking about L4-L7 layers maybe( application ptotocols ? or at least some of them.. ), I was talking about packing data, serialising it to the network, remember disks now rely in network, in big clusters.. you are serialising all the time..

And really, either way, modern machines are really fast at doing these conversions, so compared to sending/receiving stuff from memory, or (orders of magnitude more expensive) sending and receiving over the network, the conversions are effectively free.

Sending and receiving stuff from memory has exactly the same cost,
in BE it fetch the MSB and store it in a MSB significant location,
in LE you fetch the LSB and store it in a LSB significant location.. no advantages here for any one..

Inside each, its the same or almost the same, with advantages to BE in finding faster if the number is positive or negative( for obvious reasons.. ).
But I was not talking about 'archs improvements', but instead about formats..

[carewolf]

Number in BE are correctly printed, in hexa, as it his a natural notation for humans.

I like the idea that hexidecimal is the most natural notations for humans..


Don't think it is true, but I like the idea

[torsionbar28]

It's not inherently more efficient - it's only better if you're serializing to a format that is natively BE. By convention, a lot of network protocols are.

Yeah, "network protocols". LOL. You mean like TCP/IP? Nobody is using that, are they?? No need to mention that obscure protocol by name, am I right?

If they were natively LE, then the shoe would be on the other foot.

And if the moon were made of cheese, we could eat it. What's your point??

[tuxd3v]

I like the idea that hexidecimal is the most natural notations for humans..

The hexadecimal notation uses numbers and letters to specify a meaning of something..
So yes its the best notation for humans, that are dealing with machines..
And its a lot better if it his written correctly has we speak, which means...from left to right..
BE - 0x0123456789

Don't think it is true, but I like the idea

Ok,
What is for you the best human readable notation( if not correct hexa )??

[Mangix]

Wonder who actually used this. In OpenWrt, all big endian MIPS targets have limited flash space.

[cybertraveler]

Is it bad practice to create new application-layer, network protocols that use LE?

[xen0n]

Is it bad practice to create new application-layer, network protocols that use LE?

IMO it's not a matter of endianness, but rather proper documentation, at which Homo sapiens is not known to be good...