There is a problem when I try to use lfconvert to convert from largefiles to normal, see: # hg lfconvert --to-normal code code2 initializing destination code2 transaction abort! rollback completed ...8<... File "/hgext/largefiles/lfcommands.py", line 116, in lfconvert _addchangeset(ui, rsrc, rdst, ctx, revmap) File "/hgext/largefiles/lfcommands.py", line 170, in _addchangeset _commitcontext(rdst, parents, ctx, dstfiles, getfilectx, revmap) File "/hgext/largefiles/lfcommands.py", line 260, in _commitcontext ret = rdst.commitctx(mctx) File "/hgext/largefiles/reposetup.py", line 284, in commitctx node = super(lfilesrepo, self).commitctx(*args, **kwargs) File "/mercurial/localrepo.py", line 1248, in commitctx fctx = ctx[f] File "/mercurial/context.py", line 1185, in __getitem__ return self.filectx(key) File "/mercurial/context.py", line 1227, in filectx return self._filectxfn(self._repo, self, path) File "lfcommands.py", line 152, in getfilectx fd = open(path, 'rb') TypeError: coercing to Unicode: need string or buffer, NoneType found The path at that line contains "None" which is not a valid path in this case. Whole process terminates. Current hg version: branch stable Mercurial Distributed SCM (version 2.2.2+27-2440822446ce+20120629) I'm available for more info, minor debugging (dont speak python too much, sorry) and testing if you'd like.
The problem is likely that a largefile isn't available in the system cache. There's a TODO comment near the point of failure asking what to do if the file isn't cached. Adding this snippet to the end of test-lfconvert.t (and setting the usercache property in the hgrc at the top of the test) demonstrates the problem, but it isn't a problem if the rm line is commented out: $ hg clone -U largefiles-repo issue3519 $ rm "${USERCACHE}"/* $ hg lfconvert --to-normal issue3519 normalized3519 It seems like aborting with a message that names the missing file and rev would be the minimally nice thing, but ideally it would go download the file. A workaround is to use 'clone --all-largefiles' to make sure everything is cached, and then convert *that* repo.
*** Bug 3629 has been marked as a duplicate of this bug. ***
Proposed fix here: http://www.selenic.com/pipermail/mercurial-devel/2012-October/045237.html
Fixed by http://selenic.com/repo/hg/rev/0fc1ce271ee6 Matt Harbison <matt_harbison@yahoo.com> largefiles: fix a traceback in lfconvert if a largefile is missing (issue3519) The largefile may be missing for various reasons, including that a remote repository was cloned without the --all-largefiles option. Therefore, it seems reasonable to attempt to download the missing files and failing that, abort and indicate the affected file and revision so the user can manually fix the problem. (please test the fix)