Andy Wingo <wi...@pobox.com> writes: > You should be able to remove the ref entirely with "git push :master" > then "git push master" -- as least the last time I mucked with these > things that's how it worked.
Ah cool, I'll try to remember that for the future. > BUT... I wouldn't bother, really. In the meantime I committed a few > things to `master' this morning, so we'd need to rebase anew. The merge > commit has no practical implications, and indeed will happen from time > to time with distributed development. And now if we fudge the history, > it could be that people's git pull invocations stop running, like your > snapshot did. > > So while if you really want to do this, we can, but my advice would be > to let the tangle drop off into history ;-) I'm very nearly persuaded by this... but I feel that I don't understand why some of the ways of looking at the merge commit show so many diffs. For example: http://git.savannah.gnu.org/cgit/guile.git/commit/?id=2f9ae9b1040e1b9339bb0bc8b0013a5346622c44 (On the other hand, gitweb - http://git.savannah.gnu.org/gitweb/?p=guile.git;a=commitdiff;h=2f9ae9b1040e1b9339bb0bc8b0013a5346622c44 - just says "Trivial merge", which is more reassuring, and I think reflects reality.) I'm imagining looking back through the history at some future time when I've forgotten what happened here. When that happens, I don't want to be misled into thinking that the merge commit made loads of changes, when in fact it didn't. Is it just that I'm misunderstanding what cgit shows? Regards, Neil
