On 14 May 2013 20:53, Reuben Thomas wrote: > > On 14 May 2013 19:10, Reuben Thomas wrote: > >> >> On 14 May 2013 18:10, Glenn Morris wrote: >> >>> Reuben Thomas wrote: >>> >>> > Could you summarize what you're suggesting now, please? I previously >>> > suggested one change, to add (setq buffer-offer-save t) to the function >>> > "remember". >>> >>> Which I said was fine, and that you should install. >>> If you think nothing more is needed, please install it and close this. >>> >> >> Thanks, will do. Thank you also for your additional suggestions: it was >> useful to have alternatives with which to compare my idea. >> > > Apologies, I think I've messed up the commit, and pushed a change with an > empty log message (finger trouble: I hit C-c C-c and committed instead of > C-c C-a to pull in the changelog message). > > Is there some way to fix this mistake? > (I tried to uncommit and recommit, but the bug number was no longer recognized. I guess also that would have rewritten the history of the public repository, not a good idea. But since a commit that refers to a non-existent bug can automatically refused, could commits with empty log messages also be automatically refused? This would especially help hackers more accustomed, like me, to git, which refuses empty commit messages by default.) -- http://rrt.sc3d.org