To me, this is really simple. I have the camera colorspace set to shoot in sRGb, and I shoot RAW+JPG (smallest, lowest quality JPG). Here's my thinking and workflow. First of all, realize that setting the colorspace to sRGB only affects the JPG - the RAW by definition includes all the data to make Adobe RAW or Color Match or whatever space you want. I do this because I'm going to use the JPGs for web preview, review, selection, client presentation, copyright registration, etc, and don't want the hassle of converting them to sRGB before I do any of those things which are mostly viewed on browsers, which are not color-managed. (Remember, its not the "internet" per se that creates the color problem, its the browsers. If you view your Adobe 1998 files on your own computer in a browser, they'll look equally bad whether the files are transmitted over the internet or come from your own harddrive).
I use the corresponding RAW file for my final production and publication because Adobe RAW has a wider gamut of colors (some of which can't be seen in a browser - hence the need to convert browser images to sRGB - but CAN be seen in Photoshop and in print.) I can also decide to take my RAW file and later convert it to sRGB from any other format - TIF or PSD - if I want to show that actual file in a browser (like my website).
Workflow: When I download my shoot files, I first split them into two folders, one for RAWs, one for JPGs. I then use a utility (in my case, ThumbsPlus) to AutoRename them all, running the same script on both folders so they still have the same filename (but different extension of course) as their counterpart - thus maintaining the duality, but with a more meaningful filename. At this point I can go in and start deleting outtakes (rejects) in pairs if I want to. Once done, I will run a batch and resize the JPGs down to something managable, like 800pixels on the long side at 72 dpi, and automatically imprint a watermark. This is the set from which I will make client presentation galleries and copyright registration files. Once the client or I decide which image to use in some fashion, I go back to the original corresponding RAW file and use Adobe ACR in PS2 to open and process it.
I know many photogs who don't bother with the JPG part of my solution, they prefer to save the card space and then use file converter like ACR to process all their RAW files into JPGs later, or they use the actual RAW files in Bridge or some other third party software to view and select from. One thing though, if they're shooting RAW, by default they're able to get their colorspace into Adobe 1998 or Color Match or even sRGB - whatever they want. If you only shoot JPGs, I don't see any point in shooting in RAW. You're already set for the web, and there's enough information to make decent prints and color seperations. If it was really critical that you have the maximum colorspace gamut, you should be shooting in RAW anyway.
I new shoot the Canon 5D, but this worked the same when I used the Nikon D2X and D70. Also, most pros I know don't use the manufacturer's conversion software, either Nikon or Canon. Most use Adobe ACR, some use Phase One.
Andy Pearlman Studio