I installed the fusefs-osx 0.7.0 gem but the sample scripts don't mount. I tried hello.rb and yamlfs.rb. The script doesn't abort, and the mount point's Finder icon changes to a mounted disk, but I can't access the contents. If I try "ls", I get "Input/output error".
Running lsof on the ruby process shows that the gem's fusefs_lib.bundle and /opt/local/lib/libfuse_ino64.2.dylib are loaded. "/dev/fuse0" is also in the list.
lsof also gives this error:
lsof: WARNING: can't stat() fusefs file system /fusefs/rb_hello
Output information may be incomplete.
assuming "dev=2d00000b" from mount table
The "mount" command shows (username elided):
ruby@fuse0 on /fusefs/rb_hello (fusefs, nodev, nosuid, synchronous, mounted by ...)
When I ctrl-C the script, I have to umount the point manually.
System info:
Leopard 10.5.8
ruby enterprise 1.8.7-2011.03 (i386 only)
libfuse 2.7.3 installed via Macports
I also tried installing the gem on the system ruby but I get the same result.
Google didn't turn up any results about this issue.
Well, after poking around in the gem's ext directory, and looking at the lsof output of a fuse-python script, I found a workaround. Basically, I edited the extconf.rb like so:
$ diff extconf.rb.orig extconf.rb
3c3
< if have_library('fuse_ino64') || have_library('fuse')
---
> if have_library('fuse')
Then run:
make clean
ruby extconf.rb
make
Then copy the fusefs_lib.bundle to the lib dir. This isn't actually needed, since the gem always loads from the ext dir.
I think I read somewhere that Snow Leopard requires the ino64 version.