Yes, we are running Symantec antivirus corporate edition.  We also sometimes run
a file watcher application which checks for changes in files, kind of a dynamic
"hg status" program.  (Users claim they are not running the file watcher.)  So
if an anti virus program just happens to be opening a file because its just been
changed/created, that probably prevents it from being moved.

Moving the tmp... file to dirstate effectively repairs the repository.

bramsay: Do you run any virus checkers? We already had reports about
such problems. The Windows implementation side of Mercurial seems to
be sensitive in this regard. See for example Issue1147, which reports
problems with a product named "avira" (last line in msg6124).

The revision of the working directory is stored in .hg/dirstate. If that file is
missing, the revision will appear to be 0000. One way this could happen is if we
attempt to write a new dirstate and are able to delete the current dirstate but
not rename the new one in its place.

Sorry I don't have a way to reproduce this, but I'll keep trying.  Obviously it
doesn't happen very often or we wouldn't be able to use Mercurial at all.  I
suspect this is something special these users are doing.  For now I'm going to
look at the code to see what could cause Mercurial to think its back at revision 0.

Setting to need-eg

I can't really reproduce this:

D:\tmp>hg clone http://selenic.com/repo/hello
destination directory: hello
requesting all changes
adding changesets
adding manifests
adding file changes
added 2 changesets with 2 changes to 2 files
2 files updated, 0 files merged, 0 files removed, 0 files unresolved

D:\tmp>notepad hello\Makefile

D:\tmp>mv hello newhello

D:\tmp>cd newhello

D:\tmp\newhello>hg st
M Makefile

D:\tmp\newhello>hg ci
'vi' is not recognized as an internal or external command,
operable program or batch file.
transaction abort!
rollback completed
abort: edit failed: vi exited with status 1

D:\tmp\newhello>hg ci -m "Crap"

D:\tmp\newhello>hg st

D:\tmp\newhello>hg version
Mercurial Distributed SCM (version fd91bf1585cb+20080213)

Copyright (C) 2005-2007 Matt Mackall <mpm@selenic.com> and others
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

This is on Mercurial 1.0.1.

User does the following on Windows XP.

   hg clone https://repository... myver
   (edits some files)
   mv myver build2222
   (more editing)
   hg status
   (shows some files as "M"odified and other as "?", which is normal)
   hg commit
   (response is "nothing changed")
   (At this point "hg status" shows all files in the source tree as "?", 
   i.e. not tracked.  The source tree is intact.)
   ("hg id" shows 000000000000)
   ("hg log shows the apparently complete repository, about 1000 changesets)
   ("hg verify" reports no problems)
   (there is a file named tmpcjx7c in the .hg directory.)

This is happened to two users three times.  Other users seem to be working fine.
 The repository is 1.8Gb.