Handling differences in file encoding between code streams for the same file

Is there any best practices document describing how to avoid encoding problems between windows and Linux developers working on the same files. What we're experiencing is that files in the say the Release 1 code stream is not UTF-8, and then say in the Release 2 code steam, the file is UTF-8. When trying to apply changesets with no differences besides the encoding, RTC gives very odd errors such as the one shown below. Needless to say, this is absolutely MADDENING to deal with this issue when merging.

Is there a way to go through our entire code repository and normalize the file encoding of text files (maybe those within a set of file suffixes)? Any way to set a preference in RTC (using Server v4 and client v3)? Any other tips for merging? We also have issues with differing line terminators, but that issue is not as onerous as the encoding issue.

Thanks,

Justin