Hmm, good arguments on both sides. The whole "realism" says; why not have some people building buildings, while others produce troops, all in the same town, but some people feel "game play" is better when you're forced to make hard decisions between one or the other.
I can think of a compromise or two;
A) Make it an option at the start of the game (one building queue or two). This is by far the hardest I would think, because the AI would be required to play intelligently both ways, so there would be a lot of effort involved with this decision)
Bring back the two separate production lines, but make there be a production penalty if both are running at once (like James009D mentioned). If you really, really needed a building or some troops fast, you could hit a "pause" button on the other type of production, to get what you needed produced at full speed again. It still makes for interesting choices, and to be honest it really does feel a lot more realistic.
C) Keep one queue, but somehow keep track of what a city has been producing, either buildings or troops, and give a small but accumulating bonus to the OTHER type of production, that will go away once something of the other sort has been built. The bonus shouldn't appear after creating either one building or one unit, but rather after perhaps 3 either bildings or units in a row.
Say for example that city A has produced 4 different buildings in a row without building any troops. It could now have a + 20 % to production for the next troops it produces. If the player build another building instead, the bonus could grow up to 30 %.
This would represent the fact that even though you haven't REALLY been building any troops, the city has been "training" some guys in the background. If it was done the other way (producing a lot of troops, and now you have a bonus to make a building), it could represent that there was some work being done on the building "in the background" while the city was focused on producing troops.
Anyway, these are just some of my ideas. I really like C, but i don't think it would work well in this game, it would probably work a lot better in a much more complex game that really focuses on economics (like some of the WWII games I have played, can't remember the names right now).
Yes, I realize that all of these ideas would make programming the AI to play intelligently even harder, because the AI would have to be taught how to intelligently take advantage of the system.
Also, I definitely second the idea of being able to reorganize the queue. It's a very important function, for lots of reasons. It's one of those polish things; if that ability isn't present, I know somebody somewhere was being lazy.
<edit: Whoops, idea "B" is a chillin dude.>