a...@gnu.org (Alfred M. Szmidt) writes: > ChangeLog entries are trivial and quick to write, and save so much > time in the future. The small repetiveness is a insignificant price > to pay for the benefit of the ChangeLog file in various forms.
Repetiveness, even if small, is still a bad thing when a computer can do the task for you. ChangeLogs: - are error prone as an human needs to write them. In fact, we needed to invent a way to amend ChangeLogs when we generate them from the git log (gitlog-to-changelog has a way to amend old ChangeLog entries). - make merges painful. Here as well we needed to invent a tool to deal with this issue (git-merge-changelog in gnulib). - assume someone is going to "undo the changes" without any help from the computer. While it is trivial to do this using a DVCS. - force projects that use gitlog-to-changelog to have a quite unuseful log since it is maintained in the changelog format. As was already pointed out in this conversation we miss the *why* in favour of *what* changes were done. Cloning a repository is the de facto way of distributing source code nowadays. If there are better practices, I really see no reasons to not embrace them. Maintaining ChangeLogs is a waste of time both for the developers writing them and for whoever is going to use it. Regards, Giuseppe