I don’t see a problem mixing those two OSs if that’s what you really want to do. If we’re honest, it’s pretty edge case stuff we are talking about and most users experienced enough to be mixing Windows and Linux systems would be aware of the limitations of Windows when it comes to naming things. Granted the reserved namespace problem, COM1 et al, is far less discoverable; but it’s also far less common of a trap. So it’s not something that comes up that often if your team are aware of the problems.
Worst scenario is you’re dealing with a non-technical team but in those instances you’d keep your storage centralised and use an NT based solution (or something with first party support for AD) so you catch any Windows related bugs at save time rather than further down the road when you need to access them again. Given file names in Windows is limited by both the file system and the OS, whereas Linux is limited only by the file system, all you really need to do is catch those Windows-related errors. So running an NT-based storage array would solve that problem.
