So I wanted to copy all of my User data onto a newly formatted HFSX partition, whilst preserving file creation times, owernship, and so forth. The problem is that using the built-in Mac OS X rsync tool resulted in a writefd_unbuffered error, and in any case, would just copy the ._ files faithfully to the new HFSX drive, which would then not be picked up by Spotlight (since it would probably look in the resource fork itself).
In the end, the easiest way to do it was to use the zip utility. Running it with -X -y -r was enough to preserve extended information (-X), as well as symlinks (-y) and recursively (-r). When unzipping it onto an HFSX drive, the ._ files are seamlessly merged with the meta information, which is then Spotlight-indexable. So, after a tortuous route copying files, formatting the drive, and copying back again, I can now start installing Mac OS X Server! Read on for a few things to note...
A few points worth noting:
- I've not verified what happens when an existing file has resource forks. Either it will be merged sensibly, or the resource data stored in the ._ will get preferential treatment, or vice versa.
- The zip files don't seem to preserve ownership IDs, though they do preserve the read, write, and execute bits. Actually, they might work, but since my OpenDirectoy is down, it can't resolve the UIDs anyway, so it may be showing my user by default
- I had a few problems copying files, with zip saying I/O error: invalid argument (input file read failure). Some -- but not all -- of these I managed to resolve by deleting the corresponding ._ file. In other cases, I used the folder action 'Make archive,' which runs the zip file as well, though I've yet to verify whether that stores the symlinks or extended attributes
The plus side is with all this copying, Spotlight should have no excuse for not indexing the entire set of user content that's going on the hard drive.

