From a12f3f58abae3172c943debce370c11091906334 Mon Sep 17 00:00:00 2001 From: Joey Hess Date: Tue, 11 Jan 2022 13:07:51 -0400 Subject: [PATCH] comment --- ..._5ad0596373cc363ec9bc0b69efbac03c._comment | 21 +++++++++++++++++++ 1 file changed, 21 insertions(+) create mode 100644 doc/tips/using_signed_git_commits/comment_4_5ad0596373cc363ec9bc0b69efbac03c._comment diff --git a/doc/tips/using_signed_git_commits/comment_4_5ad0596373cc363ec9bc0b69efbac03c._comment b/doc/tips/using_signed_git_commits/comment_4_5ad0596373cc363ec9bc0b69efbac03c._comment new file mode 100644 index 0000000000..7ac72d40d7 --- /dev/null +++ b/doc/tips/using_signed_git_commits/comment_4_5ad0596373cc363ec9bc0b69efbac03c._comment @@ -0,0 +1,21 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 4""" + date="2022-01-11T17:02:59Z" + content=""" +Are we still concerned about this? Well, git has a workaround for SHA1's +insecurity and will eventually change hashes. There are plenty of other +reasons to want to sign git commits, certianly. + +The webapp bypasses gpg signing because it commits automatically and +potentially frequently, and depending on how gpg handles password +prompting, that could flood the user with repeated password prompts. +But you can change this default with the `annex.allowsign` configuration. + +(Commits to the git-annex branch are also not signed by default, for similar +reasons. Also, the risks of SHA1 collisions involving the git-annex branch +seem small to nonexistant, since that branch only records bookeeping +information git-annex cares about, and a small amount of configuration. +git-annex does not use data from that branch in any way that would let +an untrusted person who modified the branch do anything malicious.) +"""]]