Po Lu writes: > Manuel Giraud writes: > >> Alright. But in such case, I do not really know what to say: I am >> adding one argument to all callers based on its siblings' values. > > In this (and only this) case, our practice is to omit functions modified > solely to introduce the new argument from the log message, and to append > to the entry designating the recipient of the new argument the words > ``All callers changed.'' > > So long as the default value for the argument is straightforward, of > course. If computing the argument demands non-trivial changes, then the > ChangeLog should relate them as it would any other. What about this patch then?