I continue to be amazed at how antiquated ideas continue to get in the way of getting good output.
The overall concepts are extremely simple. All digital color, especially large format, is about managing a conversion from what the color is in the file to accurate output on your specific print system. To do this at a reasonably good level, you simply connect the color space of the file with the output profile that matches your output in your RIP's color management settings. Done.
If your file is RGB, CMYK, Pantone, or a mix of these, it should not matter. If your system is setup correctly, and you have a good output profile, life should be pretty good. The details of this can get a bit complicated, but if you have the proper goal in mind, it can guide you when you are confronted with choices.
I'm going to generalize here, so try not to pick apart exceptions.....
For most folks, your design applications should be setup to recognize and create files in these color spaces....
RGB = sRGB
CMYK = US Web Coated SWOP (in the US)
If all your design apps Working Spaces and your RIP's input profiles are setup to agree to these settings, you've got your input side pretty much sorted out for 99% of files. Now you'll be able to handle all of these file types with equal ease and success.
The output side is generally the hardest part to do well because many "canned" profiles are of average performance. Only you can define if they are good enough for your needs or not. But most people have no idea how to evaluate and test the output profile objectively.
Sending perfect grays though the print system as even numbers of RGB or as CMYK numbers of K only or all CMY and K mixed, but mixed as a properly balanced grey, will be a very demanding test of an output profile. If the resulting print does not come out grey, that's a clue the output profile is inaccurate for your system. Another valid test is to send Pantone colors though with the RIP's ability to recognize them correctly turned ON. Then check for accuracy. If the Pantone is defined as a standardly named Pantone color, and your RIP "sees" it correctly, this is also a valid test.
If you need better results than your valid tests give you, expect to get into making your own profiles. Then your color performance is only limited by the equipment, materials, and your expertise at getting the most out of them. For some shops, this is a monumental change of the better. For some, the advantage is not worth the expense and frustration. Usually the differences are purely economical and obvious.
Of course there is the issue of out of gamut colors. Having a Pantone Bridge book is an excellent tool to help visually explain to print buyers why they can not get Reflex Blue or Orange 021 C on a print system using the CMYK that is defined in the Pantone book. But a properly profiled print system that uses the typical OEM inks can hit 80-90% of the Pantone book with little effort or editing on a decent material. Your level of success with Pantone most often has to do with the accuracy of the output profile, not gamut limitations.
Some common exceptions to these very simple guidelines are if you create the large majority of what you print. If so, going with Adobe RGB as a day to day working space is a good choice. The RIP obviously needs to be setup to agree with this choice. You also have to be aware that almost 100% of RGB files you get from outside sources are going to be sRGB. This creates an opportunity for your RIP to think the incoming file is Adobe RGB instead of correctly agreeing with the incoming sRGB.
Sometimes it is helpful to turn proper color management OFF in the RIP to run CMYK files with the printers maximum gamut instead of striving for color accuracy. This has its place, but will cause more grief if all your CMYK files are run this way. To do this workflow well, your RIP's linearization (calibration, not icc profile) has to really ROCK! Generally, stick to converting all incoming files (assuming your output profile does at least an average job of converting them).
I don't intend to fan any fames here as so often happens with color discussions on this board. I hope to illustrate that a properly setup print system is not hard to achieve. If you want to get the last 10-15% of performance out of your prints, expect it to get more complicated (and expensive).