|This is the talk page for discussing improvements to the POSIX article.
This is not a forum for general discussion of the article's subject.
|This article is of interest to the following WikiProjects:|
- 1 Single UNIX Specification
- 2 POSIX compatibility
- 3 POSIX subsets
- 4 Stallman's intended expansion of "POS"
- 5 Why the cleanup tag?
- 6 Other OS'es
- 7 Is Windows really compliant?
- 8 Linux
- 9 Linux and Windows
- 10 Microsoft Windows Services for UNIX is not an operating system
- 11 UNIX System Services
- 12 Link need update or removal
- 13 Inter-article conflict?
- 14 po6
- 15 reference #8
- 16 MKS Toolkit
- 17 Is Minix POSIX as stated - I very much doubt it?
- 18 Controversies
- 19 Mac OS X and POSIX
- 20 expansion of acronym
- 21 General Meaning
- 22 Unix 03 certification versus marketing
- 23 What is most current
- 24 POSIX and G.R.O.S.S.
- 25 Stallman
- 26 IEEEIX/POSIX source
- 27 FUD
- 28 Tandem Computers
- 29 Unwarranted removals.
- 30 BSD sockets as "POSIX"
I don't think it is a good idea to merge the POSIX and Single UNIX Specification articles. Judging by how many articles link to POSIX, it is a very current and wanted topic. note that even the article Single_UNIX_Specification itself links to POSIX ;). Even though POSIX is beeing "phased out" in favor of Single UNIX Specification, that doesn't mean that the article has to be too. Compare XFree86 and X.Org, virtually all Linux distros today are using the X.Org implementation, yet the article XFree86 still exists and should keep existing. 22.214.171.124 00:09, 9 Apr 2005 (UTC)
- Donald Axel writes:
- No don't mix Posix and Single Unix. There are differences though mostly legal, as far as I know: Posix is ISO and is mandatory by DOD for any computer application system. (By ISO is meant a standard acknowledged and recommended by ISO or more precisely by International Standards Organization/International Electrotechnical Commission, ISO/IEC subcommittee 22.).
- I would like to dig more into that question but do not have resources now. Google: POSIX "Portable Operating System Interface" "Cross Platform" yields (nr.2) Microsoft Solution Guide for Migrating High Performance Computing (HPC) Applications from UNIX to Windows
- Additional: I don't know where to get the full story on POSIX and RMStallman, but I suspect that IX is the "Unix" hint and X is cross platform. RMS is notorious for multi-purpose acronyms.
- At the moment I do not have access to ISO-2001-2002-2003, but will look up comments on the name as soon as possible.--d-axel (talk) 21:57, 31 July 2011 (UTC)
- IX: lots of unices ended in IX (e.g. dynix, ultrix, minix, xenix); POS: Piece Of Shit. --moof 08:29, 10 January 2006 (UTC)
- My humble opinion: saying that the POSIX article should be merged with Single UNIX Specification is like saying that because Bush now has Clinton's function, the article on Bill Clinton should be merged into George W. Bush.
- Had the SUS been made, maintained and officially accepted as the successor to POSIX by the IEEE, then a merger would be more justified, but even then, they are two separate things. Heck, if we just said "oh, they're similar, let's just merge them!" to every article with remote connection to another one, I bet we could reduce all of Wikipedia to a single article. --Sir Link 22:03, 28 October 2006 (UTC)
- After about half a decade, I would like to revive the discussion about the relation between SUS and POSIX. Both articles currently talk about POSIX:2001, POSIX:2004, and POSIX:2008, therefore adding unnecessary redundancy to Wikipedia. Referencing the Bush/Clinton argument: Sure, if there is historical significance to keep both the SUS and the POSIX article, then both articles should be kept. But we should add explicit information in both articles about how these two terms relate to each other. By the way: who is Donald Axel? --Abdull (talk) 19:19, 9 March 2010 (UTC)
Donald Axel is a sysadm/lecturer and I have been member of a local ISO organization (ISO is rather business oriented, not so free, earns money by selling specifications and so on, you buy a membership or if you can give competent insight you get an invitation). --d-axel (talk) 21:57, 31 July 2011 (UTC)
Which OS'es are certified as POSIX compatible? I only know of Windows NT, are there others?
- I wasn't aware that NT was POSIX compatible. But mainly the UNIX-like OSes fit into that area. Things like Linux, BSD, Solaris, etc. As for certification, the article links to this, which may clarify. MichaelBillington 00:54, 23 July 2006 (UTC)
This part of the article needs some serious help. Saying something is POSIX compliant is a legal statement and most of the OSes listed are not certified. The article was missing 3 big ones here, AIX, HPUX, and Solaris. If you take a look at the list of compliant OSes maintained by IEEE and the Open Group you will not find BSD, OS X, or Windows. I think this part of the article is pretty misleading, especially when you put FreeBSD, Linux, etc as "aiming for compliance".
- I fully agree. Some fact checking might be nice on this article, people. —The preceding unsigned comment was added by 126.96.36.199 (talk) 12:34, 28 December 2006 (UTC).
- On the page it lists Mac OS X as fully compliant, but I don't think it is because the semaphores part of the pthreads library isn't implemented on it. I'm not 100% sure if that's a compliance issue; but everywhere I look says that pthreads is part of the POSIX standard.
- This situation is not much better as of September 2007; Saying that Linux is "mostly compliant" is misleading also, since, for all intents and purposes, it is POSIX compliant, it's just that there are many flavours and no one distribution is going to pay IEEE's fees for certification. Some elaboration on this point would be useful and informative. --Rhombus 16:13, 10 September 2007 (UTC)
- Mac OS X is not POSIX compliant for many reasons. I typically forget about the issues after a while but any release that did only offer BASH-3.x cannot be POSIX compliant and I recently discovered that Mac OS X does not fill in most members of the sructure siginfo_t used by waitid(). This makes Mac OS X pretty unusable for software that blindly assumes POSIX compatibility. Schily (talk) 15:30, 22 June 2015 (UTC)
- The section about "mostly compliant" still looks rather arbitrary. E.g. I cannot think of anything in Contiki that is POSIX compliant. — Preceding unsigned comment added by 188.8.131.52 (talk) 14:33, 8 August 2016 (UTC)
OpenVMS from Digital Equipment (now HP) is POSIX compliant. POSIX interface was implemented around 1990 starting from posix 1003.1 and going on with the others interfaces (shell and utilities, real time...) --Mimmetta 11:32, 26 September 2007 (UTC)
- The question would be: to what revision of the standard is OpenVMS compatible? Schily (talk) 15:30, 22 June 2015 (UTC)
This should include the different subsets of POSIX. For instance, POSIX 2003 is the newest one, and only IBM AIX passes that one. Stuff like that should be mentioned. Someone who knows more about these things should add to the article!
Stallman's intended expansion of "POS"
From the article: "The term POSIX was suggested by Richard Stallman (not as a compliment) .... POSIX has been backronymed to Portable Operating System Interface". Is this implying Stallman expanded POS as "piece of shit"? If so, the article should say so explicitly. Indefatigable
- I wondered the same thing. It isn't very clear the way the article explains it. Jdufresne 23:17, 24 February 2006 (UTC)
- At the risk of saying "me too".. I will. If Stallman did intend for POS to stand for piece of shit, I'd like to know his reasons and have them at least referenced by the article. If not, then the article should be reworded. Workaphobia 04:35, 13 September 2006 (UTC)
- Uh.. or POSitive? Isn't HURD intended to be UNIX esque (and therefore presumably, POSIX compliant in its design) Hmm..? —Preceding unsigned comment added by 184.108.40.206 (talk) 00:22, 11 March 2009 (UTC)
Why the cleanup tag?
I don't see why it's there. Other than the acronyms this page seems fine to me. Anyone care to explain? MichaelBillington 00:47, 23 July 2006 (UTC)
- I removed it. I think there is a vast conspiracy to make Wikipedia look like crap by adding such tags to many articles. :P --Pmsyyz 23:52, 15 September 2006 (UTC)
None of the above are POSIX compliant. NetBSD and FreeBSD are close, but not there yet.
Isn't BeOS POSIX compliant?
-Mostly, but not entirely. A brief (and unfortunately a bit dated) explanation can be found here (see #69 and #70) --Apophos (talk) 21:10, 8 April 2008 (UTC)
penOS/blagOS - do these exist or are they just persistent vandalism?
Is Windows really compliant?
Isn't POSIX compliance a facet of the operating system itself? Does POSIX compliance really count when added by separate payware software packages? Paul 04:04, 8 October 2006 (UTC)
Windows is POSIX 1003.1 compliant since Windows NT. After Windows NT this compliance shall be activated using SFU 3.5. Whatever this is free or not , has nothing to do with compliance. SFU 3.5 is neithertheless free if you have windows.
- it's still provided through an addon of some sort, so windows should be in the "Compliant via compatibility feature" section, no?
Michael, with all due respect, the guys on freenode are hardly likely to have much of a footing in reality. Linux is not POSIX certified. See http://get.posixcertified.ieee.org/search_certprodlist.tpl?CALLER=cert_prodlist.tpl
Integrity, on the other hand, actually is, so you should put it back.
- Linux is definitely NOT POSIX compliant, even with glibc and the GNU userspace tools added. It's very simple to make Linux non-compliant, disable Unix domain sockets (it's a kernel option). If anything it might be compliant with 'additional features' and 'added tools'.
- Also interesting that someone added a link to http://standards.ieee.org/regauth/posix/ for Linux, even though it isn't on the lists there. —The preceding unsigned comment was added by 220.127.116.11 (talk) 12:31, 28 December 2006 (UTC).
It's a pity that from this article it seems like Microsoft Windows is more Unix than Linux is. I would like to propose that on the list of "POSIX-compliant" systems there must be only certified products. Additional software (patch!) which is compliant to POSIX does not mean that the system becomes compliant itself. Ask IEEE, do not make original research please! 18.104.22.168 13:12, 28 August 2007 (UTC)
Linux and Windows
Without wanting to repeat the two above discussions, but with bringing them together:
- It is strange that Windows is in a listing that appears as being more POSIX compliant than Linux
- The listings of Windows seem to have a lot of exceptions in them ("When using ...", "Include the ....") that seem to be deliberatly vague.
- If only certain editions of Vista support it, then it shouldn't say "Windows Vista" in the list, because having some editions of it be compliant doesn't make the complete "Windows Vista" POSIX compliant. —Preceding unsigned comment added by 22.214.171.124 (talk) 13:54, 28 September 2007 (UTC)
Can the split up between the listings be made different, instead of having a list "Fully Compliant" and "Mostly Compliant"? The fact that some versions of Windows appear in a list as more compliant than Linux could be a sign that the criterium on which the two lists are based isn't so well chosen.
- POSIX was created by the U.S. government (the DoD?) solely for procurement purposes. The U.S. federal procurements needed a way to legally specify the requirements for various sorts of calls for bids and contracts in a way that could be used to exclude systems to which a given existing code base would NOT be portable. So, for example, Microsoft's Windows NT was written with enough POSIX conformance to qualify for some government call for bids ... even though the POSIX subsystem was essentially useless in terms of practical portability and compatibility with UNIX® systems. But see my major rant ;-) It would be nice to find Wikipedia:RS for the first statement though. User:ScotXWt@lk 17:16, 29 June 2014 (UTC)
Microsoft Windows Services for UNIX is not an operating system
The list "Fully POSIX-compliant" under "POSIX-oriented operating systems" includes "Microsoft Windows Services for UNIX". However, Microsoft Windows Services for UNIX isn't an Operating System but a software package. Why is this included in a list of Operating Systems?
--126.96.36.199 07:31, 1 October 2007 (UTC)
UNIX System Services
Link need update or removal
The "OS Comparison" external link has outdated information on UNIX as it is since a couple of years available in a free, open source version (Solaris/OpenSolaris) which is not reflected. --Thommym (talk) 20:06, 26 November 2007 (UTC)
- I don't even think you noticing it was a coincidence. Sam Barsoom 00:47, 1 April 2008 (UTC)
PS: I forgot to quote (about *BSD systems)
They diverge from the mainstream UNIX (and POSIX) capabilities They only run on x86 processors
MKS is another popular UNIX for Windows system. I don't know how it rates against POSIX standards. I would guess pretty good, because it would be good marketing strategy. Rhmccullough (talk) 13:03, 23 June 2008 (UTC)
- The MKS toolkit is only a set of commands; it does not provide the C API of the POSIX standard.Professor Tournesol (talk) 10:13, 26 July 2009 (UTC)
Is Minix POSIX as stated - I very much doubt it?
The article says Minix is fully POSIX compliant, yet the Minux article says it is a Unix-like operating system. If if was POSIX compliant, I doubt it would be a Unix operating system would it not? I personally think Minix should be removed from the list, but will not do it until I get some feedback from others.
- You might be confusing some facts. There's no reason an OS can't be both POSIX-compliant and Unix-like. C xong (talk) 03:42, 2 June 2010 (UTC)
Is the Controversies section of any use? I can think of a gazillion things most operating systems do differently than POSIX. Does anyone use SCCS? Has anyone ever noticed Linux ps isn't exactly as what POSIX describes? —Preceding unsigned comment added by 188.8.131.52 (talk) 15:55, 22 January 2010 (UTC)
- Linux is an operating system kernel delivering an interface to disk, net and other devices; "Linux" as such is not responsible for utilities like "ps" (process select and show). The system interface to the kernel is even "hidden" behind the Gnu Libc which implements Posix but also add when needed. The good thing is that Linux projects combine to a system on which you may find (almost?) all Posix specified functionality - and then some. Putting so many projects together in a way which actually works would not be possible without a strong Unix specification.
- --d-axel (talk) 21:48, 31 July 2011 (UTC)
- I also found this strange. I don't even understand the argument. Why would 1MB blocks be "intuitive" for du? The file system block size would be intuitive because that's what du is supposed to take into account. From the manpage, it would seem the story is also a bit more complex than that paragraph would make it seem: "Display values are in units of the first available SIZE from --block-size, and the DU_BLOCK_SIZE, BLOCK_SIZE and BLOCKSIZE environment variables. Otherwise, units default to 1024 bytes (or 512 if POSIXLY_CORRECT is set)." --Mudd1 (talk) 06:01, 31 March 2016 (UTC)
Mac OS X and POSIX
Mac OS X may have passed the POSIX copliance tests but it is not POSIX compliant as it only provides bash-3.x as a shell callable via "sh". Bash-3.x is known to have major bugs (e.g. bash -e is not implemented correctly which e.g. causes makefiles to continue after errors) that make it non-POSIX.
Mac OS X also comes with GNU make that is known for being not POSIX compliant because it's Makefile parser does not handle space characters correcty and because it calls commands via "sh -c cmd" instead of "sh -ce cmd". --Schily (talk) 09:06, 14 July 2011 (UTC)
- On the other hand, the given source asserts "Full POSIX compliance and UNIX03 certification". Do you have a WP:RS source in mind, or is this merely a passing comment? TEDickey (talk) 10:53, 14 July 2011 (UTC)
- A failing SW build process as a result from missing POSIX comliance is of course a reliable source. And BTW: I believe that you are on the POSIX mailing list where we did discuss the related problems. Should you have a problem to remember this discussion? In our case, "fully certified" obviously does not mean fully POSIX compliant. --Schily (talk) 11:30, 14 July 2011 (UTC)
- You appear to be referring to a mailing list, which is not the actual deciding body. Individuals do make comments, some lead to documentation updates. From observation, it appears that the sponsors get to make the decisions. An example of WP:RS would be this TEDickey (talk) 22:43, 14 July 2011 (UTC)
- I am sorry to see that you repeatedly ignore facts. This is not how a scientific discussion works. Apple may have passed the test-suite but this just verifies that the test suite is incomplete and for this reason let's Mac OS X slip though... POSIX compliance means compliance to POSIX. Passing the test suite seems to be easier. BTW: the documents say: "Intel based system - default configuration", this leads to a configuration with a file system that is not POSIX compliant as it is not case sensitive. --Schily (talk) 09:12, 15 July 2011 (UTC)
- To put this to rest, Mac OS X version 10.6 Snow Leopard for Intel based Macintosh computers is UNIX® 03 certified. Sorry, I hate OS X just as much as anyone else but it is certified. --J Grant Fox (talk) 21:56, 7 August 2012 (UTC)
Recent edits removed WP:RS claiming that they don't exist. That includes the page from The Open Brand - Register of Certified Products for OSX 10.5, which looks (aside from date and version number) the same as the one for 10.10 (no adequate justification was given in the editor's change comment). TEDickey (talk) 23:08, 20 May 2015 (UTC)
- If that link was in the article, it would probably not have been removed.... Note however that an OS that offers bash-3.x as the only possible shell cannot be POSIX compliant, as bash-3.0 has too many bugs and POSIX deviations - regardless of what a test suite is able to find. This has been discussed in a POSIX telephone conference when bash-4.0 was upcoming and at that time, the bash maintainer could be convinced to fix the most annoying bugs by me and David Korn as a result of a long discussion. So any Mac OS X with bash-3.x that might have received an approval slipped thorough as a result of flaws in the test suite. Also note that the default filesystem in Mac OS X is not POSIX compliant and we could not find out how Apple might have configured a system to get a POSIX compliant filesystem. Schily (talk) 08:48, 21 May 2015 (UTC)
As one can see, you removed the link, and as usual you continue to argue on the basis of unverifiable private conversations rather than reliable sources of information TEDickey (talk) 23:50, 21 May 2015 (UTC)
- You two are talking at cross purposes. While you mention the difference between RS and OR over and over again, he only cares about the scientific truth. ;) 184.108.40.206 (talk) 12:03, 26 June 2015 (UTC)
expansion of acronym
The acronym is not defined in the given source. Also, contrary to the latest edit, it does not have any wording like "based on", but says:
The Austin Common Standards Revision Group (CSRG) is a joint technical working group established to develop and maintain the core open systems interfaces that are the POSIX® 1003.1 (and former 1003.2) standards, ISO/IEC 9945 parts 1 to 4, and the core of the Single UNIX Specification, Version 3. The approach to specification development is "write once, adopt everywhere", with the deliverables being a set of specifications that carry the IEEE POSIX designation, The Open Group's Technical Standard designation, and the ISO/IEC designation. The current set of specifications is simultaneously ISO/IEC 9945, IEEE Std 1003.1 and forms the core of the Single UNIX Specification Version 3 .
It would be helpful to have a paragraph or two better explaining what POSIX is in practical terms. What is it for? What does it accomplish? What makes a system that is POSIX-compliant better (or worse) than other systems? Is there a drive for all operating systems to be POSIX-compliant? Why or why not? 220.127.116.11 (talk) 22:08, 14 April 2012 (UTC)
I came to this article not only to find out what POSIX is - which it answered - but also to learn: why should a POSIX set of standards exist? Of what utility is it to have a POSIX and to have OSes be in compliance with it? Who finds it useful, and for what purposes? I didn't get that question answered. A brief paragraph about this, for dumb newbs like me, would be an excellent addition to the article.
Unix 03 certification versus marketing
- 10.8 Mountain Lion is now on the list (as of at least 18 July 2012). Strange that 10.6 Snow Leopard is present but 10.7 Lion was skipped … 18.104.22.168 (talk) 13:43, 18 July 2012 (UTC)
What is most current
POSIX and G.R.O.S.S.
- Regarding these two edits, we have a bit of a conflict. IEEE Std 1003.1 calls it "Portable Operating System Interface", but the following sources call it "Portable Operating System Interface for UNIX":
- --Guy Macon (talk) 22:11, 19 January 2014 (UTC)
- No, we should go with what the majority of reliable secondary sources say. Wikipedia:No original research#Primary, secondary and tertiary sources applies. --Guy Macon (talk) 22:51, 19 January 2014 (UTC)
"Richard Stallman suggested the name POSIX to the IEEE. The committee found it more easily pronounceable and memorable, so the committee adopted it." Found POSIX to be more memorable than what? either this whole sentence should be dropped, or expanded to actually matter IMO Bumblebritches57 (talk) 10:00, 28 February 2014 (UTC)
- "POSIX" is more memorable then IEEE Std 1003.1 like "Ethernet" is more memorable then IEEE 802.3, though sadly the Wikipedia spreads bullshit again... "Bluetooth" is more memorable then IEEE 802.15.1, and "Firewire" more memorable then IEEE 1394. Sadly IEEE 802.11 has not such "more memorable name", so people call it however they want often WLAN (does it stand for "wave LAN", or really for "Wide area network"?) , and – of course – the Wikipedia makes night shifts to spread half-knowledge and bullshit. User:ScotXWt@lk 07:24, 29 June 2014 (UTC)
Actually a third-party source is preferable to a first-person recollection. Looking for one, I found this: file Proceedings of the 38th Hawaii International Conference on System Sciences - 2005, POSIX – Inside: A case study Jim Isaak TEDickey (talk) 01:15, 5 March 2014 (UTC)
- Nice finding, thank you! By "an environment that supported portability between systems" does he mean API = source-code portability or ABI = binary portability (cf. The Linux Programming Interface)? Sadly the Wikipedia is not very keen on respecting this decisive difference, especially when referring to the Windows API, the regularly mean the Windows ABI. Now how about POSIX/SUS? User:ScotXWt@lk 07:33, 29 June 2014 (UTC)
Whether the Wikipedia is used to spread and sow FUD (Fear, uncertainty and doubt) or whether there are just not enough people with the necessary knowledge willing to contribute, or both.
- POSIX was created by the U.S. government (or the DOD) solely for procurement purposes; it is NOT an interface designed from scratch in a clean way
- The Single UNIX Specification (SUS) is a set of standards for an OS to be met, to qualify for the brand-name UNIX®.
- The specification described by POSIX and SUS are identical. Cf. "POSIX.1-2008 is simultaneously IEEE Std 1003.1™-2008 and The Open Group Technical Standard Base Specifications, Issue 7" The difference between POSIX and SUS are legal: The Open Group owns the brand-name UNIX®, to get it, you have to certify against SUS, not against, POSIX, since POSIX is a standard specified the IEEE...
- UNIX® is a registered trademark, a brand. It is used for certified operating systems compliant with the Single UNIX Specification.
- Unix-like is a term to describe operating system, that probably would comply with the Single UNIX Specification to be allowed to advertise themselves with the brand-name UNIX, but nobody payed the money to actually make sure they really do... While I personally don't mind people calling Debian, Fedora or Ubuntu "unix-like", the article Unix-like is far to long, if not superfluous. The article Linux is again total crap, its introduction should contain something like
- The term "Linux" describes the family of Linux kernel-based operating systems, not any one operating system... the base components, such as the Linux kernel (its SCI, system call interface), the glibc, the GNU Core Utilities and some others, make many Linux operating systems "unix-like", i.e. = no money to get certified against SUS and actually get UNIX® brand-name.
- the term "Linux-based" is braindead. It is either Linux kernel-based or simply "a Linux", another one.
First somebody has to know all of the above. Then he needs to gather references. Then he needs to do contribute to the Wikipedia and write articles, that reflect these facts in a clear language. BUT then, he needs to fight all the hard-working "Wikipedia-people" (let's not call them "contributors", since they don't contribute, they are more like, we are around, and stir the shit) that either willingly spread FUD or are simply fucking stupid. User:ScotXWt@lk 08:39, 29 June 2014 (UTC)
Open standards compelled a billion dollar company to adopt a POSIX environment:
"In 1994, the NonStop Kernel was extended with a Unix-like POSIX environment called Open System Services."  Xtian (talk) 11:48, 17 October 2014 (UTC)
Seven OSes (incl. QNX, BSD/OS, A/UX) have been removed even though reliable sources proved they had been certified. (e.g. http://web.archive.org/web/20120726162344/http://get.posixcertified.ieee.org/cert_prodlist.tpl) They should go back, perhaps with a note mentioning that they are not currently certified. Agree/disagree?--Elvey(t•c) 02:09, 5 June 2015 (UTC)
- BSD/OS and A/UX most likely have never been certified as they are very old. Note that before SUSv2, the certification was a joke as there have been successful attempts to prevent important parameters from appearing in the POSIX standard and as a result, passing the certification was not worth anything. This is why governmental sites asked for the SUS as an enhancement standard. If you can send reliable sources for something that fully passed at least SUSv2, you are welcome. Note that your URL verifies that e.g. Integrity did not pass a full test and thus is only partially POSIX compliant. You see, sources need to be read carefully. Schily (talk) 12:42, 8 June 2015 (UTC)
BSD sockets as "POSIX"
An editor appears to equate BSD sockets with POSIX. Actually, POSIX is not synonymous. For start, read the POSIX documentation. Whether Winsock is POSIX-compliant is not addressed by the comments in the FAQ; a suitable source has not been provided. TEDickey (talk) 00:17, 9 October 2015 (UTC)
- Sure, BSD sockets and POSIX sockets are not synonymous; however, the former is an earlier revision of the API on which both POSIX and Windows sockets are based. Whereas POSIX mostly adopted BSD socket interface as is (with extensive additions and only minor alterations like header files), the sources make it clear that Winsock has significantly diverged from BSD sockets, and that these differences remain (despite, one might argue, adopting some later POSIX additions like getaddrinfo() and other IPv6-enabled functions/structs). Therefore the cited sources are adequate (if not the best suitable) to support the claim that Winsock is not POSIX-compliant. 22.214.171.124 (talk) 02:03, 9 October 2015 (UTC)
- The argument lacks substance. The statement is original research as it is written and sourced. TEDickey (talk) 07:57, 9 October 2015 (UTC)
- TBH I don't see this statement as controversial, as any socket programmer (and given the sources, any competent reader) should be able to verify its truth, even if the statement does (strictly speaking) constitute a synthesis—assuming the distinction between BSD and POSIX sockets is practically necessary here (which it isn't, in my opinion). So to me, the statement "Winsock does not conform to POSIX" falls into "The Pope is Catholic" category. But perhaps it's just me; do you actually challenge this statement and mean to argue that Winsock is POSIX-compliant, or that controversy exists about its POSIX compliance status? 126.96.36.199 (talk) 12:18, 9 October 2015 (UTC)
- Actually, any competent programmer should be able to provide a suitable source, since this is the sort of detail-oriented thing that programmers are adept at. Combining sources to state something not explicitly in the sources is (besides being outside the guidelines), counter to the premise of competence. If you find a competent programmer, they'll have to agree. TEDickey (talk) 23:32, 9 October 2015 (UTC)
- Unfortunately, competent programmers' suitable sources (besides manuals, standards pubs, and occasionally textbooks) include experience, knowledge, and source-code-sharing resources generally not accepted by Wikipedia as reliable, such as stackoverflow, github, habrahabr, etc. Like so: . Then, there are university and training courses, conferences, etc, where the attribution of a particular statement to a particular source may not be verifiable at all. Most programmers I know are practical people, they are not keen on excessive pedantry as it's counterproductive in a work environment, so personal experience (and that of colleagues) usually counts as a suitable source, as long as it's repeatable/verifiable. They are also engineers, which means combining various sources of knowledge is anything but contrary to competence; in fact this is a prerequisite for it. All this is probably beside the point here (I only mention it because you brought it up).
- My other point still stands: given the sources, the conclusion only counts as synthesis if we actually need to distinguish BSD and POSIX sockets for the purpose of the statement, which (all things considered) I don't think we do. If we don't, then the sentence "There are a limited number of instances where Windows Sockets has diverted from strict adherence to the Berkeley conventions" or "In two cases it was necessary to rename functions that are used in Berkeley Sockets in order to avoid clashes with other Microsoft Windows API functions" is about as clear a statement of non-compliance as we can expect from Microsoft. Additionally, I am yet to hear anything, sourced or unsourced, which challenges the statement itself (as opposed to its attribution). So I suggest we leave the statement as is until a competent programmer joins the discussion does so. If you are still unhappy about equating BSD sockets with POSIX sockets here, we could change "POSIX-compliant implementations" to read "Berkeley-based socket implementations", but IMO this would only serve to confuse the reader. 188.8.131.52 (talk) 20:37, 10 October 2015 (UTC)
- That's a lot of words which in summary attempts to put words into my mouth. TEDickey (talk) 00:28, 11 October 2015 (UTC)
- I'm looking for an actual reliable source which supports the statement you made in the topic, and have not seen any credible discussion on it so far. TEDickey (talk) 00:28, 11 October 2015 (UTC)
- The sources referenced are as reliable and appropriate as can be reasonably expected on this topic, and they are adequate to support the statement made. The "lot of words" was to show why I believe so; since you disagree, it was an attempt to get a two-way dialogue going, where I could hear out your objections and reassess my view, possibly conclude that the statement should be edited or removed. But since I never heard any relevant objections, I maintain my position. If you still disagree with the edit as unsourced/unverifiable (in Wikipedia's sense), then I suggest requesting a third opinion. 184.108.40.206 (talk) 02:42, 11 October 2015 (UTC)
- Since your proposed source relies upon personal interpretation, you might consider lining up another anonymous editor to provide your opinion. TEDickey (talk) 10:38, 11 October 2015 (UTC)
Before anyone posits a third opinion...
...Are both IPs the same person? If so, registering an account would avoid further confusion. Erpert blah, blah, blah... 04:42, 17 October 2015 (UTC)
- Yep. I was logged in through a proxy and forgot about it; sorry for the confusion. 220.127.116.11 (talk) 18:28, 17 October 2015 (UTC)
|Response to third opinion request:|
|I primarily develop for *nix platforms, so I'll opine here: I understand that anon's concern, and agree that synthesizing information into unique solutions is what you do as an engineer in general. However, it must always be remembered that this isn't a software project, this is an encyclopedia intended for the general public, and that Wikipedia has rules. Our core content guidelines are clear that information in this encyclopedia must be verified by what the project considers to be reliable sources, and that there can be no original research. While we can ignore all rules, I think at the very least before we go down that path, we should seek consensus from at least the relevant WikiProjects, but until such consensus is obtained, we should conform hard and fast to the Wikipedia content policies. Hence, for the time being, I will side with User:Tedickey. Anon, I also welcome you to bring your concerns to the relevant WikiProjects as well to seek better consensus. --KRAPENHOEFFER! TALK 01:32, 18 October 2015 (UTC)|
- Well, the general consideration of professional specifics is secondary here; the primary concern is whether the information presented actually qualifies as synthesis. So in short, you believe we do need to distinguish between POSIX (as a standard) and BSD sockets (as a convention) here, and therefore consider this information synthesis? 18.104.22.168 (talk) 18:16, 18 October 2015 (UTC)