> Surely the formal standardization process precludes an Oracle-style attack on implementors...?

Surely is a strong word. Who knows where this train is going to end up once it derails? They surely offered up permissive patent licensing, and all of those patents have expired by now anyway, but would they have bothered to sign away copyrights that they didn't think they had? Copyright lasts 70 years from the death of the author, so while the patents are all ancient history, the copyright, if such a thing exists, would certainly still be active and there would be decades of infringement to pursue.

Patents that make it into standards are still fair game for licensing. If copyright is understood to apply to APIs themselves (not just the text of the ISO documentation), then why wouldn't a similar sort of licensing practice apply?

Now I'm wondering how closely the old Windows APIs resemble OS/2.

OS/2 1.x was co-developed by IBM and Microsoft. There are some resemblances between OS/2 1.x and Windows 3.x APIs, because many of those APIs were actually designed by the same Microsoft employees – for example, most OS/2 API calls starting with Win* have a similarly named Windows 3.x API call, just without the Win* prefix. (Despite similarities like that, the APIs are incompatible; they probably would have been more compatible if it were not for IBM's influence – for example, OS/2 and Windows use different coordinate systems in their graphics API, because IBM insisted OS/2 had to use the same coordinate system as IBM's mainframe graphics software, GDDM.)

Microsoft and IBM have cross-licensing agreements allowing use of OS/2 code in Windows and vice versa. Microsoft used this to include OS/2 compatibility components in old versions of Windows NT (newer versions have removed it); likewise, IBM used it to include a copy of Windows 3.x in OS/2 to enable running Windows 3.x applications. (The agreement did not include newer software they developed after their breakup, so IBM wasn't allowed to use the Windows 95 or Windows NT code in OS/2, nor was Microsoft allowed to use OS/2 2.x and higher code in Windows.)

So, the odds of IBM trying to sue Microsoft over OS/2 APIs in Windows is zero. It would be precluded by the licensing agreement between them.

Microsoft still owns rights to OS/2, I believe this is part of why IBM cannot release OS/2's source code to third party distributions of OS/2 such as eComStation and ArcaOS

I don't think Microsoft has copyright on the whole of the OS/2 source code, just components they developed. I don't know for sure, but I think it is likely that some components, such as SOM or WPS, contain very little or no Microsoft-developed code. New features developed after OS/2 2.x were developed solely by IBM, without Microsoft involvement – Microsoft was only involved in OS/2 1.x (and possibly earlier stages of 2.x development.) There might also be some components which IBM licensed from third parties (non-Microsoft), but they are likely just individual DLLs or drivers, easily separated from the rest of the code.

I think the real problem here is IBM, not Microsoft or any other third-party. There are significant sections of the OS/2 code which are only under IBM's copyright, and there is nothing legally stopping IBM from supplying the source code to those sections, but it doesn't want to. Also, regarding the sections co-owned by Microsoft, I wonder what the story is – did IBM ask Microsoft and get told "No" (or "Yes" but only under non-viable conditions)? Or did IBM never even bother to ask Microsoft about it? I have no idea, but my gut tells me the later may be more likely than the former.