So overall what this series does is to create a struct that contains config parsing state (roughly speaking...it's either parsing state or a "kvi" that contains some of the parsing state that existed when its value was parsed), starts a pattern in config.c in which that struct is passed between functions as much as possible, and establishes a few functions like reader_origin_type() that make it convenient to work on the basis of this struct as opposed to on the config_source and kvi individually. And following this pattern, I think, likely would have enabled us to avoid the bug in patch 7/8 and enable us to avoid similar bugs in the future, so overall I'm happy with this series. "Glen Choo via GitGitGadget" <gitgitgadget@xxxxxxxxx> writes: > Introduce 8/8 to get rid of the confusing acronym "struct config_source > cf", but I don't mind ejecting it if it's too much churn. I think patch 8/8 is not worth including, so I didn't leave comments on it, but I'm fine if it ends up merged too.
