document branches better
[mharc.git] / README.FSF
... / ...
1Note: This repo is prone to being force pushed because, commits are
2meant as documentation of whats changed from upstream and whats been
3deployed for significant periods, so sometimes its better to edit them
4as they are developed.
6We do not use use the provided read-mail to process spools and we don't
7use lists.def. We bypass that part of mharc and call web-archive
10This repo is based on mharc 0.7.3 release. The .dist files are all the
11upstream except changing path names of standard utilities to match the
12distro. I dunno what those are about.
14The .dist files should never be changed except when upgrading to new
15upstream version (and there does exist a non-release snapshot we should
16upgrade to).
18Any file with a correspond .in version should never be manually edited,
19instead generated from the .in file by running make configure. Note,
20make configure must be run from /home/mharc (checkout into that
21directory on your local system).
23Ansible pulls a branch with the name of the current hostname, so
24is an alias to master, and is an alias to
25This was achieved by doing for example:
26 $
28sudo -i -u git
29cd /home/git/repositories/mharc.git
30git symbolic-ref refs/heads/ refs/heads/
32The branch should be rebased to any changes from master, like this:
34git checkout
35git rebase master
36# if there are merge conflicts, manually merge any .in files, then
37make configure
38git add .
39git rebase --continue
40git push -f
42Then on
44su mharc
46git fetch
47git reset --hard origin/
50To update existing lists headers, the only way namazu supports is to
51rebuild the index, but that takes about 3 days. Instead, I put a random
52unique string into the header where the list name goes, copied it to
53/tmp, su mharc, cd to the html dir, then ran this:
55for x in */NMZ.head; do sed "s/gachJefyeonJoa/${x%/*}/g" /tmp/NMZ.head>$x || { echo failed on $x; break; } ; done
57This process can be repeated for more changes.