diff options
| author | Sean Whitton <spwhitton@spwhitton.name> | 2022-11-29 12:31:54 -0700 |
|---|---|---|
| committer | Sean Whitton <spwhitton@spwhitton.name> | 2022-11-30 14:20:59 -0700 |
| commit | 981f8fc5b6dc8ffc4deb94b6243ec6c7cac06249 (patch) | |
| tree | 05f021eeb756b3f798a2cee198ac3cf858990621 /lib-src | |
| parent | a991c88e7ae39d7c94c8733e4bc0d36099b0a626 (diff) | |
| download | dotfiles-981f8fc5b6dc8ffc4deb94b6243ec6c7cac06249.tar.gz | |
make 'mr pull' for ~/src/emacs able to update from other branches
Diffstat (limited to 'lib-src')
| -rw-r--r-- | lib-src/mr/config | 28 |
1 files changed, 17 insertions, 11 deletions
diff --git a/lib-src/mr/config b/lib-src/mr/config index 48e5e0d7..1cf3fa82 100644 --- a/lib-src/mr/config +++ b/lib-src/mr/config @@ -335,7 +335,7 @@ skip = lazy # # "$(hostname -s)" branch: # -# Rebased onto sv.gnu.org's master branch. It has +# Rebased onto a branch from sv.gnu.org, usually master. It has # # - bug fixes, reversions etc., (cleaned up versions of) which have not # yet made it onto sv.gnu.org's master, but which are required for my @@ -345,8 +345,11 @@ skip = lazy # work, at least where that doesn't require long-lived feature branches. # # A key advantage of using these development builds is that references to -# files in *Help* buffers will be to ~/src/emacs, not to somewhere owned by -# root. +# files in *Help* buffers will be to ~/src/emacs, not to somewhere owned +# by root. Another reason for having a branch for localhost is to keep +# track of what upstream commit we want to be running on this host, and +# which upstream branch we want to rebase onto, e.g. emacs-29 vs. master, +# or some feature branch we want to test here. # # This branch doesn't usually need to be pushed anywhere, but if we do # need to push it somewhere for backup purposes, we execute @@ -357,9 +360,9 @@ skip = lazy # # "athena/unstable" branch: # -# Merges from sv.gnu.org's master branch only. .deb packages produced from -# this branch are available from <http://silentflame.com/debian>. There are -# two purposes: +# Merges from sv.gnu.org's master branch or the latest emacs-NN release +# branch, only. .deb packages produced from this branch are available from +# <http://silentflame.com/debian>. There are two purposes: # # - providing byte-compiled Lisp /usr/share/emacs-snapshot/site-lisp/elpa # to builds from the "$(hostname -s)" branch @@ -369,9 +372,9 @@ skip = lazy # my current WIP. # # Updated only when new fixes, features or reversions on sv.gnu.org's -# master branch are wanted on machines other than my development laptop. -# To do that, first "melete" is updated using "mr pull", and then I run -# that build for a day or so. Then +# master branch or the latest emacs-NN branch are wanted on machines other +# than my development laptop. To do that, first "melete" is updated using +# "mr pull", and then I run that build for a day or so. Then # # % cd ~/src/emacs-snapshot # % debian/merge-snapshot @@ -437,9 +440,12 @@ fixups = # avoid accidentally pushing personal commits to Savannah (push.default cannot # help here). So for convenient updates of "$(hostname -s)" to its conceptual # upstream, we have this definition. "pull" not "update" because we want to -# update from upstream master deliberately, not as part of automated updates. +# update from upstream deliberately, not as part of automated updates. pull = git checkout "$hostname" \ - && git pull --rebase --autostash origin master + && git pull --rebase --autostash origin \ + "${1:-$(git reflog $hostname -1 --pretty='%gs' \ + --grep-reflog='^pull --rebase --autostash origin ' \ + | awk '{print $5}')}" skip = lazy || ( [ "$1" = pull ] \ && [ -z "$(git for-each-ref '[r]efs/heads/'$hostname)" ] ) |