-> Well, if this really is a script error, it's not really the script, but the wordpress XML dump, referring to a
-> possible malformed or invalid unicode character in the dump file. This is what I can gather from other scripts.
-> I'll be checking my dump file shortly.
-
->> This is only part of the problem... I'm not exactly sure what's going on, and it's get late/early for me....
-
->>> I used --force for fast-import, but then everything seems deleted, so you end up doing a reset, checkout, add, *then* commit.
->>> Seems really odd. I edited the script however, maybe this is why... this is my changes:
-
- -print "data %d" % len(data)
- +print "data %d merge refs/heads/%s\ f" % (len(data), branch)
-
->>> That control character is a ^q^0 in emacs, see git fast-import --help for more info.
->>> I'll be trying an import *without* that change, to see what happens.
-
->>>> (5 minutes later)
->>>> Removing it makes it behave sanely. heh. Learned something new :). So ignore the comment just above, except for the --force part. You have to do that because fast-import assumes a clean uninitialized space.
+> It works fine.... The script is picky about having everything in proper UTF-8, **and** proper XML and HTML escaping. You need that to have a successful import. I let Emacs remove DOS line endings, and it works OK (if on *nix of some sort, of course). Thing is with this `git fast-import`, is that you have to `git reset` afterwards, (let's say you put them in posts/) `git checkout posts`, `git add posts`, then commit. I don't know if this a characteristic with `git fast-import`, but this is the way I get my posts to exist on the filesystem. If I don't do this, then I lose the data. If you get that "Not updating..." error, then just --force the import in. --[[users/simonraven]]