Talk:Windows API

Microsoft Windows: Computing Top‑importance

	This article is within the scope of WikiProject Microsoft Windows, a collaborative effort to improve the coverage of Microsoft Windows on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.Microsoft WindowsWikipedia:WikiProject Microsoft WindowsTemplate:WikiProject Microsoft WindowsMicrosoft Windows
Top	This article has been rated as Top-importance on the project's importance scale.
	This article is supported by WikiProject Computing (assessed as High-importance).

Windows.pas was nominated for deletion. The discussion was closed on 01 December 2013 with a consensus to merge. Its contents were merged into Windows API. The original page is now a redirect to this page. For the contribution history and old versions of the redirected article, please see its history; for its talk page, see here.

There is no such thing as Win64

There is not a distinct "major version" of the Windows API called Win64. 64-bit editions of Windows implement Win32. It's the same interface. Even the "Getting Ready for 64-bit Windows" article cited to support the notion that there is a Win64 is part of the Win32 documentation set (and it never says "Win64").

The reason Win32 is distinct from Win16 is that the API changed in incompatible ways. For starters, Win16 was the API for an OS that used cooperative multitasking, but Win32 used pre-emptive multitasking, so there were swaths of new functionality (e.g., synchronization objects, virtual memory), fundamental changes to the input and event models, reorganization of the fledgling Windows Registry, and the introduction of long file names. These changes had repercussions that required redesign to portions of the API. Some functions (notably many that dealt with window messages) had to be called with different parameters. To ease porting from Win16 to Win32, message-crackers and the Win32s overlay were offered to paper over some of the breaking changes. But most program required some changes to work with the new API.

For 64-bit systems, the API did not need to change in incompatible ways. The _ABI_ is different (pointer sizes, calling conventions, etc.). Details of the implementations may be different. But the _API_ is the same. The same functions accept the same parameters and return values have the same meaning.

If a Windows program cannot be compiled for both 32- and 64-bit systems, it's due to a non-portable assumption baked into the code about pointer sizes or ranges of representable values—it's not because of an incompatible difference in the Windows API. Aidtopia (talk) 14:58, 13 April 2024 (UTC)[reply]

The article says that "Both 32-bit and 64-bit versions of an application can be compiled from one codebase, although some older API functions have been deprecated, and some of the API functions that were deprecated in Win32 were removed." The disappearance of deprecated-in-Win32 functions might constitute an API change, although a citation for the claim that some APIs were removed should be provided. Guy Harris (talk) 22:21, 13 April 2024 (UTC)[reply]

API or rather ABI?

An API serves source code portability, it results in being able to compile, then execute programs. The same API can be maintained over multiple instruction sets, e.g. x86, ARM and Power Architecture. An ABI server binary portability, it results in being able to execute already compiled programs. It is not possible to have the same ABI over different Instruction sets!

This is absolutely not the same. It is much harder to implement and maintain an ABI. In case you want to run already compiled programs, Adobe Photoshop or some Video game, you care about the ABI. ScotXW (talk) 12:02, 27 February 2014 (UTC)[reply]

Internet Explorer Integration

"Internet Explorer has been an integrated component of the operating system since Windows 98, and provides web related services to applications [10]. The integration will stop with Windows Vista."

This seems to contradict Microsoft's publications — Preceding unsigned comment added by ItsProgrammable (talk • contribs) 21:45, 25 November 2006 (UTC)[reply]

Windows API not property of Microsoft

"Windows API is basically considered not the property of Microsoft as is its implementation."

What does this mean? — Preceding unsigned comment added by 217.158.203.79 (talk) 12:47, 8 May 2003 (UTC)[reply]

Better now? -- Tim Starling 01:43 26 May 2003 (UTC)

ExitWindows

This "function" is actually a macro that calls the ExitWindowsEx function. I propose removing it from the list to avoid confusion early on. We could replace it with a function that is more commonly used, as this function is not very commonly used by applications developers anyway. SteveS 22:07, 29 Jan 2005 (UTC)

History

The history section here does not make much sense grammatically and needs expansion and cleanup. Dsav 06:41, 28 Apr 2005 (UTC)

Opinion, much?

"However, in general, the interface remained fairly consistent, and an old Windows 1.0 application will still look familiar to a programmer who is used to the modern Windows API.[14]" I love how if you look at reference 14, it's a book written in 1998. Yes, clearly modern (2009) programmers will think Windows 1.0 applications look "familiar." —Preceding unsigned comment added by 173.9.45.237 (talk) 20:33, 29 July 2009 (UTC)[reply]

clean up?

For such a widely used api, this page seems to be a bit barren. I'm going to see how much I can clean it up in a few weeks, because I can see why the cleanup tag was added. I certainly wouldn't mind it if anyone was looking over my shoulder while I do so, since I am rather new to wikipedia writing :).--Codemonkey 09:49, 19 August 2005 (UTC)[reply]

I completely rewrote the history section. I still have to add a paragraph on the big win16->win32 move and the introduction of a 64-bit version of the API. Overall though, I've tried to err on the side of too much information, so some of it may need to be trimmed down a bit eventually. Any comments, if anyone is reading this, would be appreciated. --Codemonkey 16:42, 27 August 2005 (UTC)[reply]

If nobody objects, I'm going to remove the cleanup tag in a couple of days. --Codemonkey 18:43, 29 August 2005 (UTC)[reply]

I've gone ahead and removed the cleanup tag. I'm not quite finished with this article yet, but I think it has been cleaned up enough to warrant removal of the tag. If anyone disagrees, feel free to put the cleanup tag back up again. --Codemonkey 17:20, 4 September 2005 (UTC)[reply]

Hey I tried clarifying a small thing in the introduction for beginners who want to learn the Win32 API but don't quite understand what it is. 198.82.127.155 04:09, 28 August 2007 (UTC)[reply]

To do

I've had a short to do list for a while on my userpage on what I still wanted to do on this article. I haven't had too much time to do it, so I've moved it over to this page, and expanded it a bit. Feel free to add to the list, or make comments if I listed something stupid on it. Also, for reference, I got this reply on my userpage todo list: User talk:Codemonkey#WinAPI_compiler_support, from User:tyomitch. --Codemonkey 20:15, 8 October 2005 (UTC)[reply]

.NET

Doesn't the .NET winFX unnecessarily complicate it? Pretty much the only thing they have in common are that they are "an" API and run on "Windows" (and the latter even is not fully decided). "The windows API" is the native one. The .NET api's are different ones, I'd keep that apart. — Preceding unsigned comment added by 84.35.41.43 (talk) 20:30, 25 April 2006 (UTC)[reply]

Yeah, I'd agree. The "Windows API" to me means the C/C++ and maybe the VB APIs, but .NET is entirely new and is no more "Windows" than Java is. The only parts that are really Win32 specific are S.W.F and some of the low level OS semantics. Maybe that section should just be dropped, or rewritten to point to the main .net article?

Yes, I also agree. I just removed a paragraph about Mono (mentioned as Win API implementation in error). .NET is (like Java or Mono) a platform independent virtual machine and object-oriented application framework. See, Windows API has nothing to do with a virtual machine at all and doesn't encapsulate functionality into objects. .NET is on the other hand not mandatory depending on Windows API. Of course it uses native API of the host OS, but that may be an API under Linux or Mac (as soon as .NET supports those systems), too. Furthermore, .NET does not allow to do all you can do with the Windows API, because some of the Windows API functions are (of course) not platform independend and also considered "unsafe". This diskussion shows there is confusion about Windows API and .NET. So in my opinion it would make sense to clearly state how different they are. .NET is no "Windows API Version 2". --82.83.68.136 21:10, 8 August 2006 (UTC)[reply]

As there is great demand on mentioning .NET in the article, I put a note about .NET into the "Wrapper Libraries" section. .NET wrapps many of the Windows API functions, like all application frameworks and programming languages under windows do. Maybe (someday) Microsoft will merge the Windows API into .NET, but that's not relevant today. Maybe Microsoft also will merge the Native API into .NET someday, if mankind persists on earth long enough. No need to diskuss that in this article. ;) -- 213.169.107.51 09:10, 10 August 2006 (UTC)[reply]

Other implementations

I removed the citation needed tag from this section - it *is* generally accepted (I have worked on the leading reimplementation of Win32 for years), although if you wanted the legal precedence for this you would have to dig out the Sega case where it was shown that header files cannot be copyrighted.

More specifically, you cannot legally copyright an interface, only an implementation (which the Sega dispute formalised in case law). Therefore because Win32 is an interface (usually but not always paired with the Windows implementation) you can reimplement it legally.

Again I am not sure what you could cite to prove this in the case of Win32 specifically. It's just how copyright law works.

MikeHearn 11:13, 4 September 2006 (UTC)[reply]

IMimeMessage merge

I don't think IMimeMessage should be merged here; instead, I think the IMimeMessage article should be deleted. -- Mikeblas 16:45, 21 November 2006 (UTC)[reply]

See bug 9660 - this article shows a bug in Mediawiki

http://bugzilla.wikimedia.org/show_bug.cgi?id=9660

Ta bu shi da yu 06:52, 22 April 2007 (UTC)[reply]

That's not a bug. The colon after the title, before the reference, is the delineation between the <dd> and <dt> tags. I've moved the colons so that it looks a bit better. -/- Warren 07:10, 22 April 2007 (UTC)[reply]

Is Win32 still a trademark?

If so, shouldn't this be mentioned in the article, since Win32 redirects here? -andy 194.138.39.140 16:28, 10 October 2007 (UTC)[reply]

Where can we download it?

I am a beginner learning C++.

I think it would be important to include a link or some kind of direction as to where this API's files can be found so people like me can start learning it.

Where do I get the file(s) for Windows API? Is it a secret? No web site I can find seems to know.... —Preceding unsigned comment added by 71.37.65.160 (talk) 23:49, 27 December 2007 (UTC)[reply]

according to the article, they are in built in to windows. Just find a compiler that supports them i think. that is normally the case with API's i believe. 12:01, 7 May 2008 (UTC) —Preceding unsigned comment added by Oxinabox (talk • contribs)

The API is built right into Windows, so if you have Windows, you already got it. But to have compilers you need, you need the headers and associated tools that describe the API to the compiler. This is available as a part of the Windows SDK ( [1] ). You can use that with the Visual C++ line of compilers and Microsoft Visual Studio IDE (including VC++ Express [2]). This blog post explains how to use Windows SDK with VC++ Express 2008. The Windows SDK, however, won't allow you to use MinGW or GCC compilers. Use the MinGW Windows runtime/api for an analogous functionality. --soum ^talk 12:29, 7 May 2008 (UTC)[reply]

API articles

These individual API/dll subsections need articles of their own, and should also be organized in part according to implementation (in/outside of kernel, "nt exec", some service, system library, etc). anyone have any ideas for how to do this, or votive to start an article? 68.110.212.162 (talk) 06:17, 18 February 2010 (UTC) not an ip -- not signed in[reply]

No any date per whole article.

Per whole article, no one date.
When did it appear, in 1985, together with the first Windows release ?
1981-1983 ?
"History" paragraph : what has Petzold said about 150 lines of code, still needs clarification. At least good that there is explicitly said, that he has told about Windows 1.0 SDK.

Alexander Ilyin 20:16, 4 July 2010 (UTC) —Preceding unsigned comment added by Cantregistermynick (talk • contribs)

Compiler support section

The compiler support section says that "For a long time, the Microsoft Visual Studio family of compilers and tools and Borland's compilers were the only tools that could provide this" (meaning the ability to handle Win32 headers etc.).

In my recollection, this is not true. Before Visual Studio came out, there was Microsoft C/C++, but also Zortech C/C++, Watcom C++ and Turbo C++. Zortech C++ became Symantec C++, was dropped by them, and then bought back by its original author and survives to this day as Digital Mars C++. Watcom was withdrawn from the market before satisfactorily implementing the ARM-level language. Turbo C++ became Borland C++. I owned all three. Watcom and Symantec supplied "memory extenders" which allowed Win32 programs to run on Win16 with Microsoft's redistridutable "win32s" shim DLL. I can't remember why I switched from Borland to Watcom but I think it was because Borland lacked 32-bit support.

I think you'll find the Zortech compiler was the first usable C++ on Win32/Win16, before even Microsoft had one.

Towards the end of this era COM appeared and assumed a certain binary vtable layout. At least Zortech tracked this change. I don't accept that COM is part of Win32, by the way. — Preceding unsigned comment added by 118.209.147.103 (talk) 07:54, 7 September 2011 (UTC)[reply]

The MSFT-NSA conspiracy theory

The person that added the stuff about the NSA being able to 'access our computers' should really read: http://www.dailykos.com/story/2013/06/15/1216509/-The-MSFT-NSA-conspiracy-theory --212.121.112.33 (talk) 14:02, 14 July 2013 (UTC)[reply]

Note #3 claims that Pascal was used to implement Windows

The Pascal calling convention was used because it saves on code size. That doesn't mean that the Pascal language was used. This fallacious claim should have a citation. — Preceding unsigned comment added by 50.159.105.93 (talk) 18:44, 20 June 2015 (UTC)[reply]

I'll bet Raymond Chen knows whether it is true. -steve — Preceding unsigned comment added by Stevebroshar (talk • contribs) 11:56, 29 January 2024 (UTC)[reply]

Infobox

I've added an infobox. I realize there are 2 things that may seem odd about it, so I thought I would explain:

I've used a screenshot of ReactOS, as a screenshot of Windows would not be acceptable on Wikimedia Commons. ReactOS, the 3 Windows applications, and the novel Frankenstein can all be licensed under the GPL without issue.
I saw a place in the software infobox for OS. Currently, I've placed Windows and OS/2 there as these were the only operating systems that were publicly claiming Windows compatibility. ReactOS, Wine, etc. strive for it, but do not have Microsoft's stamp of approval. OS/2 is debatable as it implemented the Windows API up to a certain version. ReactOS could possibly go there, if they explicitly claim to support the Windows API but I think they state something slightly different.

Rjjiii (talk) 05:45, 5 January 2023 (UTC)[reply]

Function categories

The overview section currently lists categories of functions in the API, but I have two concerns about that. 1) Does not seem a good purpose for wikipedia to cover all the technical details of an API. Let the Microsoft docs cover the technical details and let this article cover higher-level concepts -- mostly what's _not_ covered by Microsoft. 2) There's a note saying the list is out-of-date. What would one update to? the latest version? This article is about _all_ versions; not just the latest. Could list the categories for _each_ version, but that's too hard to do and to get right.

Suggestion: KISS -- remove the list of categories Stevebroshar (talk) 11:22, 29 January 2024 (UTC)[reply]

Half measure: Added "notable" so that there is no need to be complete. Stevebroshar (talk) 13:14, 18 October 2024 (UTC)[reply]

communication among different Windows applications

The paragraph with "communication among different Windows applications" seems to be about stuff that Windows API is not. Suggest: remove it. Stevebroshar (talk) 11:28, 29 January 2024 (UTC)[reply]

I moved it to the section on related tech. Stevebroshar (talk) 13:13, 18 October 2024 (UTC)[reply]

wrappers

Most of the tech listed as wrappers are not what I'd call a wrapper. IMO a wrapper exposes roughly the same capabilities as the wrapped tech -- just in a different form (interface). Just because one tech uses another tech, does not make it a wrapper. Consuming is not wrapping.

Wrapping requires design intent to replace all consumption need for the underlying tech. For example, MFC and ATL are not intended to replace all consumer need for Windows API.

Taking a step back: What is the point of this section? For example, it says MFC uses Windows API. ok. Everything built for Windows uses Windows API. Goes without saying. Adds no value. What is interesting about MFC WRT WinAPI?

Suggest: Either remove this section or don't say the other techs are wrappers -- just list them as other Microsoft Windows tech that the reader might want to peruse. Stevebroshar (talk) 11:54, 29 January 2024 (UTC)[reply]

I updated that section (some time ago). No comments on the change or to my comment here. Hope it's OK. Stevebroshar (talk) 13:12, 18 October 2024 (UTC)[reply]

History

The history section starts with The Windows API has always exposed a large part of the underlying structure of the Windows systems to programmers. This had the advantage of giving them much flexibility and power over their applications, but also creates great responsibility in how applications handle various low-level, sometimes tedious, operations that are associated with a graphical user interface.

I don't know what do with that, but it's got issues.

WinAPI exposes functionality; not structure
WinAPI does not force developers to handle low-level stuff; it allows them to do low-level stuff
How is GUI related to this?

Stevebroshar (talk) 13:25, 18 October 2024 (UTC)[reply]

"WinAPI exposes functionality; not [the underlying structure of the Windows systems]" Given that a lot of the Win32 API was implemented in Windows 9x, the structure of which is a bit different from that of Windows NT, it definitely doesn't expose the "underlying structure" of Windows.

"WinAPI does not force developers to handle low-level stuff; it allows them to do low-level stuff" Perhaps what they meant is that if you use only the Windows API, you have to do the low-level stuff yourself.

"How is GUI related to this?" Perhaps they're saying that doing the low-level GUI stuff is more work than doing it elsewhere. But a citation for all of that would probably be called for here. Guy Harris (talk) 00:15, 14 November 2024 (UTC)[reply]