There have been many rumors that the name “OS/2” was chosen only shortly before the product was announced. It’s not entirely clear what the name would have been otherwise, but quite likely DOS 5; certainly DOS in some form.
There are various vestiges of the old naming; for example the OS/2 system call interface is implemented in a library named DOSCALLS.DLL—not OS2CALLS.DLL as one might expect. (It’s actually called DOSCALL1.DLL on disk for some unknown reason, although the internal module name is DOSCALLS, and the corresponding import library is DOSCALLS.LIB.) In the IBM C/2 compiler, version 1.0, libraries are named in the format SLIBC3.LIB and SLIBC5.LIB, where “3” means DOS 3, i.e. plain DOS, while “5” means DOS 5, i.e. OS/2. Later Microsoft compilers used “R” for real-mode DOS and “P” for protected-mode OS/2 libraries.
The OS/2 Museum recently scanned and converted into PDF a pre-release version of the Microsoft OS/2 1.0 Setup Guide. This document was published in April 1987, a quarter of a century ago, very soon after OS/2 was officially announced. It was shipped as part of the famous $3,000 OS/2 SDK in May 1987. The interesting fact about this manual is that except for the title page and preface, it nowhere mentions OS/2. Instead, it consistently talks about DOS.
It is to be expected that the manual would be unfinished. There are no screenshots because the look of the OS wasn’t yet finalized. After all, this was about eight months before the announced release date. The layout is somewhat poorly chosen and ring binder holes cut into the text in the table of contents. Again, such cosmetic issues are to be expected.
However, the fact that the manual only talks about DOS is not expected. At first glance, one might be confused into thinking that it really is a manual for DOS 3.3 or thereabouts. On closer look, that isn’t the case. The manual talks about CONFIG.SYS settings like THREADS, IOPL, or SWAPPATH. It explains the difference about real-mode and protected-mode environments. Device drivers that never existed on DOS are documented, such as COM.SYS. In other words, it is indeed an OS/2 setup guide.
The manual gives some idea of how complete OS/2 was in April 1987. As it turns out, the kernel and base drivers hadn’t changed much between then and the 1.0 release. OS/2 was already quite usable (and no doubt used by the developers at Microsoft), if a bit rough around the edges.
All in all, the manual is an interesting piece of PC computing history. Certainly fit for a museum.