Talk:OS/2

OS/2 FAQ

I have found [this usenet tread] about this articel whit contains some comments that possibly can be used to improve this articel. giskart 23:14 Oct 17, 2002 (UTC)
Response: <wickipedia-reply@paperlessconscience.com> We clearly need to clarify our positions, the OS/2 FAQ (deemed credible because it was written by an IBM employee) states that 3.1 was an old update, and that Windows 95 code-name 'Chicago' would not compete with it.

http://www.mit.edu:8001/activities/os2/faq/os2faq0106.html

Clearly, IBM released 3.0 without Microsoft, and (also according to the same source) eventually offered to use an existing copy of Windows for compatibility. This suggests that the Microsoft relationship was tenatious, and that they eventually denied OS/2 the prospect of full compatibility.

Why am I being cautious? Because the usenet thread is vauge, and the e-mail addresses (some @home.com) are dead. Sam

Edit details:

changed the time things started to fall apart to 1.3, since this is when IBM took over fully, according to the FAQ. The Usenet link agrees with me.
An encyclopedia should never (IMO) put things in its articles that are debatable.. so I removed "being 32-bit" and added "eventually supported 32-bit APIs" . My guess is that the usenet poster was failing to distingush between emulators, and the native OS application architecture. In other words, OS/2 supported 16bit applications, but was not 16-bit. [somebody has since edited the article to clarify versioning, stating that 2.0 was 32-bit. This is fine. Thank you.]
added links to websites I looked at, and a note that Windows 3.1 may not have been a response to OS/2.

I would like to add that while 2.0 was 32 bit some subsystems, notably device drivers, have remained 16 bit. Also 2.0+ fully supported the 16 api from 1.x Dryeo 04:45, 13 November 2006 (UTC)[reply]

ESR

ESR has a good resume of history of OS/2 in The Art of Unix Programming Davidme

HPFS is accessible through NT4

Due my MSCE 4.0 course back in 1998, we learned (and proved in practice) that Windows NT4 can indeed READ HPFS partitions (but not write them, however).

Can you give a source for this? AlistairMcMillan 15:23, 5 Nov 2004 (UTC)

It's common knowledge. Ask anyone with OS/2 skills. It can also be done with Windows 2000. Not sure abut XP though. Tannin

Alistair, I can give you the source: Any Microsoft Press Book related to Windows NT4 MCP/MCSE. However, I think I see were the confusion comes from. Windows NT 4 can read/detect HPFS partitions, but you can't install WindowsNT 4 on a HPFS partition. As Tannin says, it is common knowledge.

I wasn't disputing the fact. I just wanted to see a source. We are supposed to be dealing in supportable facts here, not personal recollections. AlistairMcMillan 13:15, 9 Nov 2004 (UTC)

I've made a change. Correct me if I got anything wrong. AlistairMcMillan 13:23, 9 Nov 2004 (UTC)

Windows NT 3.xx can be installed, convert to NTFS, read and write (but not format), and check (but not defrag) HPFS partitions.

Windows NT 4 onward lacks UHPFS.DLL (for checking and converting) and PINBALL.SYS (for reading and writing), so it is unable to do anything with HPFS partitions.

However, you can use the DLL and SYS from Windows NT 3 manually installing it. From 2000 onward the IFS changed so much that they are fully incompatible.

— Claunia 12:41, 3 September 2006 (UTC)[reply]

Here is an article on reading HPFS partitions in Windows NT 4, but you need the NT 3.5x CD. --ozzmosis 13:02, 3 September 2006 (UTC)[reply]

Windows 2K does allow the addition of UHPFS.DLL and PINBALL.SYS for reading and writing HPFS partitions. There is a package on the web that automatically installs HPFS support on Win2k. This is most likely illegal so no link. Unluckily PINBALL.SYS does not properly support large partitions (4.3GB+?) giving errors about chkdsk needing to be run when attempting to traverse directory trees. It also has a bad habit of marking partitions dirty forcing OS/2 to run chkdsk on startup. Chkdsk is very slow on modern large partitions. It is Windows XP that no longer supports OS/2 at all. Dryeo 05:39, 13 November 2006 (UTC)[reply]

Shadows

The feature I remember most in my brief encounter with OS/2 was "shadows", somewhat like MacOs aliases. Can somebody confirm/deny/annotate the appropriate articles? Thanks!

No problem. Let's work backwards. You are failiar with shortcuts in Windows? Shadows were the thing that MS copied for the shortcut idea, but copied rather incompetently. (Whether IBM hd copied the idea in turn from something else, I don't know. It's a reasonable assumption though.) With a shadow, you can do everything that you can do with a shortcut, but you can also perform any operation except delete, and have it reflected in a matching change to the real object. Change the properties of the shadow, and you change the prperties of the object itself. Much more powerful. (Successive versions of Windows seem to have gradually improved, such that Windows shortcuts in (e.g.) W2K are now approximately the same as OS/2 Shadows.) Tannin 09:03, 30 Mar 2005 (UTC)
- Ah, so we are talking about symbolic links here?--81.42.164.73 03:22, 9 September 2005 (UTC)[reply]
  - No. Shadows are not filesystem entities. Uncle G 05:26:49, 2005-09-09 (UTC)
- Also, the Shadow(s) allways know where the original is placed. So if you move the original object the Shadow will follow its trail. You can also have many Shadows pointing to the same object, each Shadow possibly placed in different folders or embedded in other objects.

This dynamic and live one-to-many relationship makes shure that the WPS doesn't have the "Broken shortcut" problem that Windows have.

Microsoft

Microsoft and this article disagree on why Microsoft switched to Windows. --Taejo 10:44, 22 Jun 2005 (UTC)

BOFH

Feels a bit like something is missing from the article without a mention to the BOFH's dislike of OS/2 users, don't it? :) --81.42.164.73 03:22, 9 September 2005 (UTC)[reply]

Agreed. I came here just to see if it was mentioned or not (I was a bit interested in finding a rational explanation for it, if it's in reference to something). I didn't seriously expect to find a mention here, though :-P 130.233.22.111 12:09, 28 May 2006 (UTC)[reply]

Yeah, it's really not encylopedic to list people who don't like something. Besides, I'm not sure there's enough room. I was sold on the hype, and really wanted to love OS/2, ran it for a while, but unlike everyone described, I just couldn't get it to be faster than Windows. And launching Windows-in-OS/2 to run a software was a pain. DigitalEnthusiast 20:17, 21 December 2006 (UTC)[reply]

IBM charging for the SDK

My memory may be rusty here, but I pretty much remember it the other way around: MS charged (a lot, if I recall) for OS/2 SDKs and then left ISVs high and dry when they decided to make windows their central strategy. Maybe someone can dig references? --Jope 13:47, 24 October 2005 (UTC)[reply]

Is Warp 5.0.1 the most recent version?

Going to http://www-306.ibm.com/software/os/warp/support/ and clicking on "All Downloads" loads to a page where one can select version "5.0.1" of OS/2. Rumors say, this version can run win32-software and is available IBM-internally only. One one knowing more? --85.180.60.183 12:24, 20 November 2005 (UTC)[reply]

Your looking at the version drop down box that is used with the "Operating Systems" (Linux/AIX etc) drop down box to the left. v4.52 is still the latest revision of OS/2 . - Cheers Ian Manners -

Quote

@During the next 10 years, millions of programmers and users will utilize this system:

Taken into consideration where NT came from... Shinobu 04:51, 22 December 2005 (UTC)[reply]

Corrections

NO full 32bit version of OS/2 ever existed, (with the exception of OS/2 for the PowerPC) particulairly the device drivers all had to be 16-bit. The only exception to this was video drivers with the very latest versions, which could finally be 32-bit, but drivers such as Networking and Storage all are 16-bit.
The free demo CD of OS/2 Warp 3.0 Red was distributed all over Europe, bundled with computer magazines. They where time limited, but installing a FixPak would remove the expiry timer, and effectively gave you a free version.
IBM has official withdrawn support
TCP/IP stack is only BSD based in the last version, older versions where based on the old IBM TCP/IP stack. This includes OS/2 Warp 3.0 (dial-up TCP/IP support only) and Warp 4.0 which had a forward port of the old IBM TCP/IP 2.0 and MPTS (Multi Protocol Transport Services). Only starting with OS/2 Warp Server for e-business (v4.5) did the BSD stack get introduced.

—the preceding unsigned comment is by 67.87.7.65 (talk • contribs) 02:35, 7 January 2006 (UTC1)

Observations and corrections

I hesitate to edit the page directly just yet, but offer these observations:

1) Device drivers can in fact be 32-bit. HPFS386 was perhaps the earliest (and most common) such driver. All device drivers in the ENTIRELY 32-bit OS/2 for the PowerPC were, obviously, 32-bit. Most low-level drivers are, however, 16-bit. Not that it actually matters.

HPFS386 is the only "first-party" (Microsoft) 32-bit filesystem driver for OS/2.

Every other filesystem driver is 16-bit. There is no documentation for making them 32-bit.

Other kinds of drivers can be 32-bit perfectly, there is documentation from IBM from OS/2 2.0 (but filesystem)

OS/2 for PowerPC doesn't use an OS/2 kernel, but a IBM microkernel, so drivers do not follow the OS/2 APIs at all and cannot be included, in any statement, with the OS/2 x86 drivers.

My two cents

—Claunia 00:06, 24 November 2006 (UTC)[reply]

2) The infamous SIQ does not stand for Single Input Queue. If it can be said to stand for anything, it would be Synchronous Input Queue. Events delivered to message queues by Presentation Manager are sent synchronously, because a very few messages (such as WM_ERASEBACKGROUND) require a reply to be meaningful. It has always been the case that a reboot was rarely necessary when an application stopped responding to events. Attempting to switch focus via ALT-ESC or CTRL-ESC brought up a dialog box asking whether to terminate the offending application. In later versions, Aynchronous Focus Change was added (it can be turned on and off at run time), which will remove focus from unresponsive applications without requiring that they be terminated - when they begin handling events again, they continue functioning normally. In rare cases, none of this works, requiring a restart of the Workplace Shell (and in rarer cases still, when it failed to restart properly, a reboot).

3) DOS and Windows executables can be launched in OS/2 commandline sessions. They are executed in a new DOS or DOS/Windows session. The popular 4DOS command interpreter has OS/2-specific functionality, including the START command, which lets OS/2 executables be launched from DOS prompts. There are also a handful of other third-party programs allowing synchronous execution of OS/2 executables from within DOS sessions.

4) While an integrated system cache for all file systems does indeed have great benefits, there are downsides as well, at least as implemented in Windows and even Linux. Moving large quantities of data in the file system around will use up all available system memory as file cache. While not fatal to new memory requests (as the cache memory is quickly freed), the peculiar decision (in both Windows and Linux) to allow memory to be swapped out to make room for more file cache means that already-running programs must often be swapped back into memory when focus is restored after a lot of file system activity. Having a fixed cache size at least prevents this nuisance.

5) Isolating Windows programs in their own Windows session is optional, not required. It's perfectly possible to run many Windows programs in a single Windows session. The ability to separate programs into separate sessions is an advantage that allows one to run, simultaneously, multiple Windows programs that don't get along well with others. An ability that did not appear in Windows NT derivatives until much later.

6) One limitation which is not mentioned is that for reasons of maintaining compatibility with 16-bit OS/2 programs (running natively), the process address space was limited to 512MB, some of which was shared memory. In practice a typical OS/2 program has between 300 and 400MB of address space. With Warp 4 FP13 and later, however, the kernel offers a newer memory allocation option that gives applications to up to 3GB of address space, with the caveat that a number of API's cannot be used with memory so allocated.

7) The TCP/IP included with OS/2 Warp was indeed for dial-up only, but there was a separate (and more expensive) version called Warp Connect which offered complete LAN support. LAN capability had always been available as an add-on to OS/2 with a separate package prior to OS/2 Warp.

Conflicting histories about OS/2

Those that win the battles usually write the histories. Although I believe that whoever wrote the article was being as straight-forward and honest as possible, it is difficult to escape some of the FUD surrounding OS/2 created by Microsoft all those years ago.

I was one of the unfortunate souls that worked on OS/2 from the IBM side. Fresh out of university, I found the OS/2 project challenging and full of potential. With the option of working with such a powerhouse as Microsoft, how could it go wrong?

It did go wrong, and it was not all IBM's fault for it to go so wrong.

Microsoft had a very active role in making sure that OS/2 would never succeed. I seriously doubt their intentions post 1989/89. I suspect IBM gave them key funding that helped them push through both Windows and NT development. NT was supposed to be OS/2 3.0! It doesn't get much more visible than this.

The overall point is that OS/2 had weaknesses that were being addressed starting in OS/2 2.0. Things like DOS and Windows support were no longer a major issue and the potential for having a fully 32-bit development environment was strong. It would not be until 1995 with Windows 95 that Microsoft would crack this open.

It turned out not so much to be about technology but rather strategy in the market. Microsoft made it next to impossible at that time to sell an alternative operating system. Bulk licensing implied that hardware vendors were paying Microsoft license fees regardless of what operating system was pre-installed. Vendors had no reason to look elsewhere.

I think the most annoying thing was that Microsoft would start the mantra of the limitations of OS/2 1.x without focusing on the possibilities of 2.0+. It made sense for them to grab the whole OS pie instead of just half. Why share when you can have it all.

I became a bit jaded after this experience and left IBM in 1993 to join a startup company called Citrix. Suprisingly, Citrix developed a successful variant of OS/2 called WinView in that same year. We found that we had more success if we avoided mentioning OS/2 and focused instead in what we could do.

IBM helped to sink OS/2 but Microsoft was largely to blame for its failure. That which could be used by Windows and NT was salvaged from initial work done on OS/2. Innovations quietly got sucked in to become Microsoft only domain. — Preceding unsigned comment added by 61.68.82.110 (talk • contribs)

Unless you can cite sources for the above, that readers can check, your personal testimony is not useful to the article. Uncle G 16:06, 22 June 2006 (UTC)[reply]
- It is pretty widely known (and discussed in the NT article) that Windows NT was developed by MS from the OS/2 3.0 project. A personal anecdote doesn't belong in the article, but it can be useful as background for editors working on it. --Blainster 21:24, 22 June 2006 (UTC)[reply]
  - No, it cannot. Personal testimony merely serves to fuel outbreaks of the operating system wars on talk pages that are meant to be for discussion of editing the encyclopaedia article. Useful background for encyclopaedia editors are sources, plain and simple. Uncle G 02:29, 23 June 2006 (UTC)[reply]
    - Yes, it can. Without saying it, the personal testimony here is implying that this page does not have a neutral point of view relating to historical events. That may or may not be true, but is worth looking into. DigitalEnthusiast 20:24, 21 December 2006 (UTC)[reply]
    - I wrote this clip. If you feel it is worth making the claims more valid I will take the time to find the sources you want. Some of these are fairly obvious (like OS/2 3.0 and the licensing) but some are harder to prove. I admit that what I wrote is a mix of fact and opinion. I wrote this because I felt no one was talking about what happened inside IBM and the frustrations everyone inside IBM OS/2 had related to MS. I could easily say that MS was just as frustrated with IBM for other reasons. It was not a very solid partnership. I just added a email I received in 1990 related to the split in development. Jeffmuir 02:59, 2 September 2006 (UTC)[reply]

As the CEO of Advanced Programming Techniques, I can testify to the severn year court battle with IBM about the tradename WARP - a section removed from the article but now added back as of today - I am also the author of "Warp" - a product that competed with IBM for hardware dollars in the mainframe market by improving performance of batch programs - It suited IBM to quash the product alonside "borrowing" the tradename to make their PC operating system product sound sexier than the "OS/2 3.0".

While my company were being drained of cash to fight IBM, the CEO of IBM was receiving an honorary knighthood for donating PC's (that they couldn't sell) to schools.ken 18:54, 24 November 2006 (UTC)[reply]

Citations

There are large sections of this article that are not cited. For instance, see the "Future" section. Contributors may read or simply assume things are true and add them to the article, but fail to cite their source. It needs some serious clean-up. If anyone has the time/interest in finding citations of some of the assertions in this article, feel free to add one or two. Otherwise, some items may be deleted for clean up. Arx Fortis 23:02, 9 July 2006 (UTC)[reply]

Windows Bashing

In parts of this article, there are many underhand comments against Microsoft's Windows operating system, some of which appear semi-malicious in nature. For example, in the Security Niche section, there is commented that OS/2 may be as vulnerable as Windows. According to the Windows stereotype, that is correct. According to fact, http://www.us-cert.gov/cas/bulletins/SB2005.html, Windows is very secure. Just though I'd point it out. --Nick2253 00:58, 5 August 2006 (UTC)[reply]

You must be kidding. Over 800 vulnerabilities discovered in one year is "very secure"? Then why did it create a $3.6 billion security market? [1] If you are trying to infer that it is secure in comparison to other OS's, you might remember that as effectively a desktop monopoly, it is the major target, and thus more vulnerable by that fact alone, even if it did have fewer security holes. --Blainster 21:22, 5 August 2006 (UTC)[reply]

Email from Microsoft about spliting development (1990)

Jeffmuir 02:45, 2 September 2006 (UTC)[reply]

I've kept this all these years. It seems a shame to not publish it here.

While working on OS/2 as a programmer, it was fairly obvious that there was tension between IBM and Microsoft.

One day, we all received the message I am including.

IBM and MS never resolved their differences after this. As developers, we knew the importance of this message. We were no longer working on OS/2 2.0 together.

20BUILD FORUM appended at 02:10:31 on 90/04/03 GMT (by OS2BNI at BCRVMPC1)

Subject 6.67 Message from RICKD at Microsoft

To clarify the previous propagation comments. Microsoft has elected to work towards a previously discussed 2/91 release plan. IBM has elected to not take Microsoft deltaed files into the IBM Build, Microsoft has elected to not bring in the IBM files into the Microsoft Build. Until management determines the development direction, each site will probably continue "one-way" builds.

At this point in time Microsoft is assuming complete ownership of all Kernel related files and will no longer be requesting exception permissions for any files in the associated directories.

Diverting Funds

There's a section that claims that IBM became concerned with Microsoft diverting funds to Windows development. I've never read anything reputable that claims this. Certainly, one could make a case that because OS/2 3.0 became NT, that this could be called "diversion", but I don't think so. IBM got full rights to Windows 3.x in the deal, and Microsoft was legally given rights to the code they'd developed for OS/2 3.0 in the breakup. Is there any credible source that can corroborate this allegation?

cite needed overkill

this article needs a lot of work, but it seems to be rife with redundant cite-needed tags. i interned and worked for IBM throughout the early and mid-1990's and can vouch for the validity of almost everything that is said. WP:OR not withstanding, it seems someone is trying to use the tags as a blunt instrument. pending no objection, i'm going to yank most of them out. /Blaxthos 10:54, 11 November 2006 (UTC)[reply]

IBM OS/2?

If it was originally a joint IBM/Microsoft project, what renders it "IBM OS/2" rather than just "OS/2"? Guy Harris 08:46, 21 November 2006 (UTC)[reply]

I agree. I have no idea why this was moved. I think it should be moved back, and at best the IBM article forwarded, but preferably deleted. -- Tmassey 19:58, 21 November 2006 (UTC)[reply]

Microsoft exited the OS/2 stage very early in the game -- they had some involvement with the 2.0 release, however by 2.1 they had abandoned OS/2 in favor of developing Windows products. IBM developed, produced, and supported the product for the majority of its life, which is probably why "OS/2" is pretty much always associated with IBM. As a side note, MS wanted nothing to do with the naming or production of OS/2, and never disputed the name. Also, the OS/2 software packages were labeled "IBM OS/2" that I remember (from 2.0 forward). As a final note, I think he was also trying to standardize the names of articles about IBM products (IBM AS/400, etc) -- i haven't checked and could be completely wrong. /Blaxthos 18:55, 24 November 2006 (UTC)[reply]

APT "controversy"

An editor has repeatedly attempted to insert this text into the article:

However this has been contested by Monaco based Software company "APT International" who claimed that they approached IBM with a performance product called Warp with a view to joint marketing, just before IBM's OS/2 dramatic change of name. APT battled IBM for through the French High Court of justice ("Le Tribunal de Grande Instance de Paris") for seven years with IBM claiming they had a prior Warp trademark with their product CAEDS. This "warp" was however merely a reference to the Warping (i.e. bending) of materials and not a trademark as IBM most certainly would have been fully aware.

Several probelms I see (each of which, individually, would prohibit inclusion):

Unsourced - No source of this information.
Original research - makes assumptions and draws conclusions.
Speculative - Speculates about what IBM "most certainly" knew, etc.
Irrelevant - That some company approached IBM with an entirely different product they called Warp and later sued IBM over does not have any signifcance to this article. The product they sold had nothing to do with OS/2 Warp -- it's simply a dispute over ownership of the name.

Removed per norm. /Blaxthos 18:50, 24 November 2006 (UTC)[reply]

Follow up - from comment left on my talk page:

I also have original copies of letters from IBM regarding our offer to them regarding our product WARP.If you want an image of the letter I will show it with the article. Theft is not trivial and it is certainly relevant to the origin of the name. IBM reputedly spent $300,000,000 marketing a product using our trademarked name.

The letters constitute original research and does not qualify as an appropriate source for inclusion in Wikipedia. However, your comment belies your non-neutral point of view -- you refer to the name "Warp" as our trademark name. Clearly you represent the company involved in the lawsuit, and it seems that your point of view is skewed by your association with them (notice your accusation of "theft" -- keep in mind we're talking about the trademark name of totally unrelated products; so you assert that you own the name "Warp"?)! It's hard to assume good faith that you are simply trying to improve the article while having a non-neutral point of view and being involved in a lawsuit regarding the subject of the article. This is not the place. /Blaxthos 20:11, 24 November 2006 (UTC)[reply]