This patch has continued in development without getting agreement on its Desirability or Design, meaning we are going to continue going back to those points until there is agreement. Posting more versions of this patch is not going to change that.
Could you please elaborate that?
I actually think the approach that has been followed is what makes open source and collaborative development work. The initial idea was based on timestamp approach which, thanks to the input of several developers, led Marco to develop LSN based checks and move forward the feature implementation.
The numbers that Marco has posted clearly show that a lot of users will benefit from this file-based approach for incremental backup through pg_basebackup.
As far as I see it, the only missing bit is the pg_restorebackup tool which is quite trivial - given the existing prototype in Python.
