On 23.02.22 02:37, Damien Miller wrote:
Implementing uploads that go via a temporary file in the client seems much more feasible as it would be subject to these considerations.
Oh I'd *wish* ... the latest supplier we added to our SFTP server and urged to do "upload temp file, then rename" got as close as doing PUT + COPY + DELETE, so we're now seriously having inotify watch for the DELETE to trigger the postprocessing of the beforeCOPYed file ... :-/
(And then there are those like the one whose client s/w throws in the towel if our server does as much as *offer* keypair auth in addition to passwords ... *so* glad that PasswordAuthentication and AuthenticationMethods can be used in Match blocks ...)
Kind regards, -- Jochen Bern Systemingenieur Binect GmbH
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ openssh-unix-dev mailing list openssh-unix-dev@xxxxxxxxxxx https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev
