Giltonio_Santos
Hero
I'm going to keep leaning back on Mike Mearls's "can we make 12 subclasses with this class idea" for my opposition to a martial gish class. I don't see the fantasy archetype for it that isn't self referential. A paladin is different from a fighter/cleric, partially because it's based on something that's a more solid concept. Whether d&d popularized that concept doesn't matter at this point.
I'd argue that Gandalf is a gish more than a wizard, so, there's at least some support for the concept of a warrior-mage in popular culture.
For the need of 12 subclasses, while I think the character concept is important, I'd prefer the outlines to be traced around the need for mechanical support. I'd first ask for a subclass that supports the defender role, one that supports damage-dealing in melee and another that supports it at range. After that, a subclass that allows someone to play a gish mainly for buffing allies/debuffing enemies. So, we'd have four, at a minimal.
After that, we can think about some concepts that are hard to pull with 5e where it is today. Some, from the top of my head: the "Jedi Knight" (the UA psychic warrior is trying to do that) or a WoW-like Death Knight. In fact, the WoW Death Knight brings not one, but at least three gish concepts that are hard to pull with current official support: a disease carrier, a lord of the undead, and an expert of ice magic. There's also space for a transmutation (self) expert and a shadow warrior (Illusion/stealth/dex-based; one could argue that the Way of the Shadow monk already covers it).
Those are 10 of the top of my head. I've spent more time writing this post than trying to envision viable subclasses for gish types and almost managed to arrive at Mearls' purposed 12 subclass requisite.