Han-Wen Nienhuys
Han-Wen Nienhuys
I've commented on the PR.
can you rephrase this without referencing your specific FS, but what general pattern you want to support? explain why returning ENOENT is not necessary (kernel issues LOOKUP to verify that...
I think the VFS has locking to prevent the race. Of course, if change the tree behind the kernel's back, things might still go wrong, and you have to have...
in general, you should always have a test that fails before your fix and succeeds with your fix.
your two pull requests seem conflicting: if you are always calling WaitMount, then you would always call Unmount after WaitMount. Since WaitMount waits for Serve to process the first request,...
can you specify more exactly what "something wrong" means? If your server is broken and in an unknown state, then is there any sense in worrying about deadlines?
I think the real problem is that the API is designed in the wrong way. The Serve loop should be started from NewServer. Possibly, it should even do WaitMount implicitly....
Let me see what kind of problems changing the API would create.
can you check with current master? I fixed a source of flakiness.
the fix is https://github.com/hanwen/go-fuse/commit/e6fa1a12fd97f29a6833d6c69d7b94438e5e3fed