|WikiProject Computing||(Rated Stub-class, Low-importance)|
_XOPEN_SOURCE feature test macro
I'm looking for detailed information about the historic values of this macro, which is still useful for portable programming. For instance, I recently discovered that on the NetBSD 5.1 operating system, I had to define not merely -D_XOPEN_SOURCE but -D_XOPEN_SOURCE=2 in order to make the declarations of the functions popen and pclose appear in the <stdio.h> header. Is this value 2 documented somewhere? All Google-able sources point only to values of 500 or greater. I.e. either _XOPEN_SOURCE is just defined, with no particular value, or it is recommended to use 500 for certain much newer features (Unix98). Thanks for any shreds of info. 184.108.40.206 (talk) 04:17, 11 April 2012 (UTC)
UNIX or UNIX-like?
Edits were made with the comment "Replaced "Unix-like" with "UNIX", as this predates much of what "Unix-like" would imply)"
I wonder if this is correct?
My memory is that there was a lot of effort to avoid using the word "UNIX" in the specifications, as UNIX was the name of a specific product, and the intent was to allow conformance to be achieved by products that were not built from the UNIX code base. In fact, many early UNIX-like operating systems were not called UNIX because of trademark issues.
- However, X/Open deals with "UNIX" rather than "Unix-like" Tedickey (talk) 23:37, 28 September 2008 (UTC)
- Where's your evidence? Reading Colin Taylor's book, I simply don't think that's true. They started out, in his words, because many computer manufacturers were turning to systems based on "UNIX, or systems derived from it and/or its interfaces". They adopted the UNIX System V Interface Definition as their starting point, but the whole thrust was standardization of interfaces, not product.
- I suppose if you can find "Unix-like" on The Open Group's website (or any reliable source related to X/Open or The Open Group), we'll have something to discuss. Tedickey (talk) 20:26, 30 September 2008 (UTC)
Sorry, which of the following are you saying:
(a) "UNIX-like" does not mean the same as "UNIX, or systems derived from it and/or its interfaces"
(b) Colin Taylor's book, "X/Open and Open Systems", published by X/Open in 1992, is not a reliable source?
Surely, not (b), so it must be (a). Is there a term you would prefer?
(Incidentally, I would think that for most organizations, the company's own website is one of the least reliable places to look for information about its history. If there is any historical information at all, it will almost invariably be highly selective. Publications produced by the company in the past are better evidence of its aims and objectives at the time, though still of course not exactly objective.)
- "UNIX-like" is a rather abused term, being applied to anything which implements a subset of POSIX. The question of how closely a system must resemble UNIX to have the term applied to it doesn't seem to be well-defined. But you're free to apply a source to support the term. (The comment about "for most organizations" is arguable since it implies that factual information can be found elsewhere, which is not always the case). Tedickey (talk) 18:28, 5 October 2008 (UTC)
I joined X/Open staff in 1989 (and never left...). I do not recall the use of the term "UNIX-like" at the time - it's a much later term. We were clear that we were building a Common Application Environment based on the SVID, but we were also very clear that conformance to the specification did not require the OS to be based on AT&T source. OSF/1 was intended originally to be free of AT&T source yet conform to XPG3. The Preface to the Portability Guide Jan 1987 (XPG2) says "In order to provide such portability, the Group defines a Common Applications Environment built on the interfaces of the UNIX operating system, as defined in the AT&T System V Interface Definition, and covering other aspects required of a comprehensive application interface". The term CAE was central to what we were doing, and if an edit is done it might make sense to include it - perhaps something like:
"More specifically, the original aim was to increase the interoperability of applications and reduce the cost of porting software by defining a Common Applications Environment (CAE), built on the interfaces of the UNIX operating system (as defined in the AT&T System V Interface Definition) and covering other aspects required of a comprehensive application interface, such as languages, data management and networking"
BTW, UNIX should be used as an adjective, not a noun, so "UNIX-like" and "derived from UNIX" are not ideal
Under Outputs it might be useful to include something on the testing and branding programs, particularly since the test code developed by X/Open is still in use for POSIX and UNIX certification