I think the choice is int vs. size_t.

I think the biggest reason to not use size_t would be if we're using -1 as a signal value somewhere.  But that could still be done with some kind of kNoTab = SIZE_MAX.

Another reason is that I think some of the underlying views objects use ints for indexes.  But that makes this more of a question of where the casts lie, not whether to have casts at all.

Given views is int based converting TabStripModel to be size_t would mean we need a bunch of casts for the code that interacts with the two. This would rapidly spiral into the need to touch a bunch of other related ui code (such as ListSelectionModel). While views is int based (which I stand behind) I see little value in doing this.

I consider it a bug that views uses ints for things our style guide says should be size_t, but I don't think it's a bug worth trying to fix.  Tons of labor for very little benefit if any.

Yeah, I agree that this isn't something worth to fix, the risk of breaking something largely outweigh the benefits. The fact that the "count" method returns an int still bothers me but I can live with this.

Thanks for the quick response anyway, I'm not familiar with the views so I was lacking some context here.