From b25ded67d9e9147922f827e8ca0620a79e0d2140 Mon Sep 17 00:00:00 2001 From: Sean Whitton Date: Fri, 26 Mar 2021 14:48:37 -0700 Subject: rework uploading data to be in terms of CONNECTION-UPLOAD Incremental updates of items of prerequisite data was not implemented, so the previous meaning of CONNECTION-UPLOAD was not achieving much. This simplifies the core implementation, should still allow connection types to implement the method to provide optimisations, and provides a way to implement copying items of prerequisite data into chroots. Signed-off-by: Sean Whitton --- doc/ideas.rst | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'doc/ideas.rst') diff --git a/doc/ideas.rst b/doc/ideas.rst index aadba1d..d32ffed 100644 --- a/doc/ideas.rst +++ b/doc/ideas.rst @@ -24,6 +24,14 @@ Connections then we can have more specific connection types which take other arguments and construct the full command. +- It might be possible to write an implementation of CONNECTION-UPLOAD for + SSH-CONNECTION which can optimise a common case. If it can see that it is + the only item in the connection chain, and there is an old version of an + item of prerequisite data to upload already on the remote side, it can move + that old version to a temporary name, rsync the new version directly to the + temporary name so that rsync can do an incremental update, and then rename + the file to the new version. + Data sources ------------ -- cgit v1.2.3