Dear all,

I just realized something bad happened to me yesterday.
After editing a remote text file with Kate, I asked Kate to save it.
It seems the operation was only *partially* successful, in that today I find only the initial ~1M (40%) of that file.
Typically, to prevent such behaviour, the file is copied first using another name, or the existing file is backed up.
Does FISH protocol work in such a way? where could I hope to find the full version of the file? (even before the modifications for that last editing sessions would be fine)

The node with the FISH client which ran Kate has been shut down already.

Did Kate save a "filename~" file? It should be in the same directory as the file which was partially saved.

No, it did not.
I disabled that function because annoying and useless.
Or so I liked to believe...
But I thought fish would put aside the original file until the transfer of the new one was successful.
I believe rsync does that, for example.

I'm not aware of the exact implementation of the Fish protocol i'm afraid, as I have always used the sftp protocol instead.

In this case, it is quite likely that it does not put aside the original file or use a temporary file prior to writing to a previously existing file. I'm not sure if sftp does this either - it is probably up to the sftp-server implementation on the server I suspect.