Recently I've been hacking on some OpenGL for an old PowerPC Mac, which means reliving the OpenGL 1.x days.
Back in the 1.x era, it seems that 1.1 was the widest common denominator.
But this got me to thinking: if you had targeted 1.1, were there any machines you'd be excluding? Or were all 1.0-capable machines eventually updated to handle 1.1 as well?
I'm guessing maybe some very early SGI hardware might have been stuck on OpenGL 1.0? Or perhaps some very early consumer 3D cards?
You might find some SGI machines in the 1990s that couldn't run OpenGL 1.1, but that would be because the owners were still running an old version of IRIX for compatibility with some strange hardware or software that would break if they updated. In which case they generally didn't want to put new stuff on those machines anyway.
Early versions of OpenGL (1.x), unlike early Direct3D, always guaranteed that your software would run - just not that it would run fast. So an older machine might fall back to software rendering if you used newer features in the API, but your program wouldn't break.
This wasn't really a problem for OpenGL 1.1, because while the API changed in a few important ways, vertex arrays and texture objects IIRC, these were not really new capabilities, just more convenient ways of doing the same thing with fewer API calls. With the hardware of that era, where OpenGL basically ran on expensive workstations, not consumer PCs/Macs, OpenGL 1.1 was almost a pure software upgrade, no new hardware.