A question about endian-ness turns out to be the wrong question

re: A question about endian-ness turns out to be the wrong question

John Doe — Wed, 02 Jan 2013 11:54:10 GMT

@jas88, "Answer: It is."

When 32 bit Windows came out, it was the only thing running 32 bit applications. That is, until Win32s and Windows XP x64.

Who knows until when will it be true that 64 bit applications are only running on a 64 bit Windows? Who knows how can we check for other combinations? Who knows what we might need to do different? Do we need to do anything different?

[But what would you do if the answer is "You are running on a version of Windows you have never heard of and know nothing about"? -Raymond]

re: A question about endian-ness turns out to be the wrong question

Steve — Wed, 02 Jan 2013 10:28:47 GMT

Testing the endianess of the server provides the evidence that the patch upgrade is worth the risk to the (otherwise) working system.

On high reliability standalone (non-internet connected) systems that have been working for years, changing software can be a very arduous (tedious) process, proving to risk averse people responsible that change is necessary.

re: A question about endian-ness turns out to be the wrong question

Matt — Mon, 31 Dec 2012 10:54:51 GMT

@jas88: WindowsRT can be big endian.

re: A question about endian-ness turns out to be the wrong question

cheong00 — Mon, 31 Dec 2012 02:01:33 GMT

I know that they should have installed the Cumulative update, but there's some MIS that I know refrain from install patches to their system, fearing it'll break their programs. Given that they haven't installed something they should have installed 3 years ago, I'm afraid that they'll go to the "code route" if someone said it's possible.

Don't laugh... at my previous company, we have a client that is a bank still doing UAT for 20+ earlier release of our company's flagship server software... Things are like dead water that won't move forward...

re: A question about endian-ness turns out to be the wrong question

Joshua — Sun, 30 Dec 2012 14:52:50 GMT

@Matteo Italia: I agree. PDP has no mercy.

@Burak KALAYCI: I refer the honourable gentleman to the solution in the article.

re: A question about endian-ness turns out to be the wrong question

Burak KALAYCI — Fri, 28 Dec 2012 13:24:18 GMT

>what seemed like a simple question

Obviously answering that simple question (on any system) in a bug free way is not trivial as it seems: the bug with the connector production code is not being able to detect this correctly.

Also it makes me wonder how such a bug survives testing, or if any were done at all.

re: A question about endian-ness turns out to be the wrong question

jas88 — Thu, 27 Dec 2012 23:30:37 GMT

I seem to recall Raymond answering the similar question of "how do I tell at runtime if my 64 bit app is running on 64 bit Windows?" (Answer: It is. On anything other than a 64 bit system, it won't run at all!); with the possible corner case of the Xbox 360, it seems "how do I tell if my Windows application is on a little-endian machine?" is another "yes, it is". I can't imagine too many Xboxes connect to SAP servers directly.

re: A question about endian-ness turns out to be the wrong question

Matt — Thu, 27 Dec 2012 19:44:46 GMT

@JM: All versions of Windows run on either a Little endian architecture or a big-endian one (or both, such as ARM).

re: A question about endian-ness turns out to be the wrong question

Veltas — Thu, 27 Dec 2012 19:41:40 GMT

What on Earth would you even do given that the server was Big or Little-Endian? Just install the update!!!

I'm not even sure why they were wondering how to detect if the system was big or little endian... it's actually quite trivial to check. Maybe they checked it themselves, were unable to get the answer they were after (because the question was wrong in the first place) and then decided to double-check by asking MS.

re: A question about endian-ness turns out to be the wrong question

JM — Thu, 27 Dec 2012 19:17:17 GMT

@Matt: you say that as if little endian and big endian are the only options. Regrettably, they're not -- although I'm pretty sure Windows has never been ported to a middle-endian architecture.