If you are adding new files, there is no reason to omit a ChangeLog entry so repoman should refuse to commit.
adding metadata should not require a changelog entry
(In reply to comment #1) > adding metadata should not require a changelog entry > Why? It would leave a trace in the ChangeLog about that there was no metadata before this point and as such no maintainer and would tell who is the maintainer.
pointless noise imo ... you're just doing: cat metadata.xml >> ChangeLog
(In reply to comment #3) > pointless noise imo ... you're just doing: > cat metadata.xml >> ChangeLog > Yes but the point of ChangeLog is to see major changes. If we don't record metadata.xml there, they would have to resort to viewcvs.
As Cardoe stated on irc, I'd rather see repoman do the entry for you rather than repoman complain because you didn't write one. Another warning or error is just gay. You want a changelog entry, just make repoman do it. You are already passing the damn message with -m for the cvs commit. No one uses a different commit message versus changelog message anyway. They just run repoman commmit -m msg echangelog msg
I don't know about you.. but I usually do: echangelog "message" repoman ci -m "message" Not the other way around. Simply the change should be, if ChangeLog isn't one of the altered files that repoman sees.. run echangelog.
(In reply to comment #5) > As Cardoe stated on irc, I'd rather see repoman do the entry for you rather > than repoman complain because you didn't write one. Both could be implemented. If you commit with -m and no ChangeLog, repoman fails / warns. Furthermore, it could have a -c switch (like sunrise-commit) that creates echangelog and commits afterward.
We will move the discussion to bug 337853 *** This bug has been marked as a duplicate of bug 337853 ***
