Little Endian Big Endian?

Nik · Post by **Nik** » Sun Oct 30, 2005 9:18 pm

Ok Mac OS X is Big Endian and Windows is little Endian so what problems to fear when doing a Filetransfer system based on PBs Network libary communicating between Mac OS X and Windows? (All Apps in PB)
bye Nik

Trond · Post by **Trond** » Mon Oct 31, 2005 11:17 am

Nothing, I think.

Post by **Dare2** » Mon Oct 31, 2005 11:22 am

I think the actual transfer won't mess with the byte sequence. So the issue, if anything, will be the "endian-ness" of the source data, as this will be retained and therefore either has to be fixed before or after transfer.

But I am guessing.

Nik · Post by **Nik** » Mon Oct 31, 2005 11:56 am

If you also have to send the length of the content i would have to make sure that for example both sides use little endain.
Right?

Post by **Fred** » Mon Oct 31, 2005 1:12 pm

Yes, it's true. I don't think it will work for now, i will fix that ASAP.

Nik · Post by **Nik** » Mon Oct 31, 2005 2:50 pm

What do you mean it ins't working now? No function for conversion or no network support

Nik · Post by **Nik** » Mon Oct 31, 2005 4:06 pm

Code: Select all

Procedure LittleToBig(Long.l)
  PokeB(@LongBig+2,PeekB(@Long+3))
  PokeB(@LongBig+3,PeekB(@Long+2))
  PokeB(@LongBig,PeekB(@Long+1))
  PokeB(@LongBig+1,PeekB(@Long))
ProcedureReturn LongBig.l
EndProcedure

Is this correct?
I also noticed that 'UNIX' is not the same as PokeS(@Address,"UNIX") Why?
And thats te point that makes us think that it should be called XINU Problem instead of NUXI Problem *confused*

djes · Post by **djes** » Mon Oct 31, 2005 7:57 pm

A little code used for a value from a file...

Code: Select all

nb.l=ReadLong()
MOV eax,nb
BSWAP eax
MOV nb,eax

Nik · Post by **Nik** » Tue Nov 01, 2005 1:22 pm

Well it musst be correct but

Code: Select all


nb.l=0
PokeS(@nb,"UNIX",4)
MOV eax,nb
BSWAP eax
MOV nb,eax 

Debug PeekS(@nb,4)

Debuges XINU and it's called the NUXI Problem so wtf

djes · Post by **djes** » Tue Nov 01, 2005 2:04 pm

Sorry, I don't understand.

Post by **Dare2** » Tue Nov 01, 2005 2:29 pm

Not 100% sure what we are getting at here, but ...

"UNIX" is 4 bytes long, respresenting here, I think, a 4 byte integer?

XINU is how a 4 byte (long) reversal would appear
NUXI is how 2x2 byte (words) reversal would appear.

Also, are you reversing all data sent one side to the other? I think (stress, think) you would only need to reverse words/longs where those were stored as words/longs in the original data.

So
>> [long]+"ABC"+[word]+[word]+[byte] - or - [1234]ABC[12][34][1]
would go across as
>> [gnol]+"ABC"+[drow]+[drow]+[byte] - or - [4321]ABC[21][43][1]

If you are sending "UNIX" and then, on fixing it on the other side, you are ending up with "XINU" then you don't need to be fixing it.

If you are ending up with "NUXI" (unfixed) on other side, need to be fixing word reversals.

If you are ending up with "XINU" (unfixed) on other side, need to be fixing long reversals.

But I am not really sure what it is that we are talking about.

What exactly are we trying to address? What are you getting on the "other side"?

Nik · Post by **Nik** » Tue Nov 01, 2005 2:50 pm

It was a theoretical Problem

Haven't got a Mac yet but was interested in it and maybe it was good because it seems like it reminded fred at something

Well maybe I can get my Hands on a Mac then I will test this Theorey in practice. But nice to know how the made this NUXI problem up. Because who the hell would stroe a 4 byte String in 2 words

Thanks a lot for your efford

djes · Post by **djes** » Tue Nov 01, 2005 3:00 pm

In theory, no. For more details:
http://en.wikipedia.org/wiki/Endian