Arthur Bodera

Yes, it usually does. I'd expect it to emit error code (and have list of consts for each possible code) along with detailed message on what went wrong.

@a5m0 If you've got any additional information on when does this occur for you, feel free to chip in here.

My reproduce code uses `/bin/mv` :\

@EldonMcGuinness what's your `.encfs` config ?

It seems that we differ in the following: - I'm on newer EncFS `1.8.1` while you're on `1.7.4` - I've got newer `nameAlg.major` of 4, while you have 3 -...

@EldonMcGuinness i'm testing everything out now. Thanks for tips. You're on the money with the additional bytes written out of order, which upsets acd_cli mount. Interestingly, with the new encfs...

That ain't fixed by a long shot. This may be one of many different use-cases where specific order of operations causes data loss. In the meantime, we can throw in...

Why not "backport" the write buffer feature as a general write-back cache for acd_cli? That'd fix problems with ecryptfs, encfs and any other applications where data is appended in small...

> Further work will be to use temp file backing rather than memory backing if individual files get too large. @bgemmill this is covered in #314 and is not ecryptfs...

LRU cache is something different to what I meant. The caching Swift FUSE does is per file handle - a process opens a file handle for writing, writes as much...