From 29ba93ba52be831c4ff7f5527710288ba73537a6 Mon Sep 17 00:00:00 2001 From: "https://www.google.com/accounts/o8/id?id=AItOawkkyBDsfOB7JZvPZ4a8F3rwv0wk6Nb9n48" Date: Fri, 22 Nov 2013 18:44:35 +0000 Subject: [PATCH] Added a comment --- ..._33a84937c87dd2406bc090a0d2969683._comment | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+) create mode 100644 doc/bugs/Can__39__t_add_a_git_repo_to_git_annex:___34__Invalid_path_repo__47__.git__47__X__34___for_many_X/comment_8_33a84937c87dd2406bc090a0d2969683._comment diff --git a/doc/bugs/Can__39__t_add_a_git_repo_to_git_annex:___34__Invalid_path_repo__47__.git__47__X__34___for_many_X/comment_8_33a84937c87dd2406bc090a0d2969683._comment b/doc/bugs/Can__39__t_add_a_git_repo_to_git_annex:___34__Invalid_path_repo__47__.git__47__X__34___for_many_X/comment_8_33a84937c87dd2406bc090a0d2969683._comment new file mode 100644 index 0000000000..df97625f79 --- /dev/null +++ b/doc/bugs/Can__39__t_add_a_git_repo_to_git_annex:___34__Invalid_path_repo__47__.git__47__X__34___for_many_X/comment_8_33a84937c87dd2406bc090a0d2969683._comment @@ -0,0 +1,30 @@ +[[!comment format=mdwn + username="https://www.google.com/accounts/o8/id?id=AItOawkkyBDsfOB7JZvPZ4a8F3rwv0wk6Nb9n48" + nickname="Abdó" + subject="comment 8" + date="2013-11-22T18:44:35Z" + content=""" +> I am not convinced at all that it makes any sense to try to sync git repositories in this way + +I agree this is not a good solution. + + +> I realize that some people drop git checkouts into dropbox and use that, but it's a fundamentally unsound thing to do, and those people are just lucky if they manage to avoid running into problems doing that. + +Well, I don't use dropbox (nor annex assistant) but I sync a lot of git repos with unison. It is not luck, it is being careful not to create conflicts. I agree this is not ideal, and I'm looking for a better way to deal with this. + + +> Also, this feature would only be used by a small number of users + +I'm not convinced of that. If done right, nested git repos inside annex could have value. + +Let's say you work with 40 git repos some private of yours, others tracking upstream, and you want to sync them across 4 machines (home, work, cloud server, backup). I imagine your preferred solution would be to configure the 4 machines as remotes on every git repo and use mr, am I right? + +This has its problems: + +1. The set of files you want in the project git repo may not be the same as the set of files you want synced across machines. For instance, I use a large software project that I compile from sources. I want to sync binaries across machines (it takes forever to compile!), but of course I don't want them commited into the project, not even as annexed files. +2. Configuring remotes for every project is tedious. What if some remote changes url? what if I want to change the path of some of the git projects? Every time I add a new repo I need to configure all the 4 remotes. This can be scripted of course, but is not my ideal solution. + +What do you think about the submodules route I proposed? I don't like submodules very much, but in this case, I think it could become a good solution. In particular it would solve 1, 2 above and be able to merge conflicting changes on the nested repos. + +"""]]