It was pointed out on some list I can't remember some while ago that commercial fs's like xfs and jfs have a huge(!) amount of testing capability behind them in the lab with facilities individual open source developers simply don't have. That discussion was only covering one of IBM or SGI, but I'd guess it holds for both.
You are implying that all open-source projects are run by individuals? Redhat at least might beg to differ... :)
Personally I would trust reiserfs, xfs and jfs but not ext3 but ext3 is out anyway. If it's not out of the box by the distro it's dead.
I'm not sure what you are saying here? I read the above as saying that ext3 isn't 'out of the box' - which it is? Just confusing wording I guess.
Various file systems are optimised for various things like small file support, nfs support, etc. I don't know enough about it, but the reiserfs site has good concise info at what reiserfs is good at and what isn't implemented yet. I guess if you really wanted to know you'd have to put the same files on all three and test with the situation which is most important to you.
As a reiserfs advocate, you might know: Does NFS work over reiser yet? I had reiser corruption once, when I foolishly NFS mounted my home partition (a reiser volume). And a quick google search should uncover plenty of benchmarks on fs comparison. Of course, they always show their benchmarkers favourite fs as being the best, but thats what you get with benchmarks... :) Daniel