Paolo Bonzini [off-list ref] writes:
I hope this is what you meant in your reply to my message.
Almost, except I wouldn't have made this part of fetch_native()
like this:
quoted hunk ↗ jump to hunk
diff --git a/git-fetch.sh b/git-fetch.sh
index b9d7a75..e2a6878 100755
--- a/git-fetch.sh
+++ b/git-fetch.sh
@@ -164,6 +164,12 @@ then
fi
fetch_native () {
+ if test "$(get_data_source $remote_nick)" = builtin; then
+ for name in $(get_remote_fetch_branches "$remote_nick"); do
+ append_fetch_head $(git-rev-parse "$name") . "$name" "$name" || exit
+ done
+ return
+ fi
eval=$(echo "$1" | git-fetch--tool parse-reflist "-")
eval "$eval"
but probably would have created another sub, fetch_fake.
I removed completely the configuration variable since it made
no sense to me, either.
I think that is sensible.
Currently the patch is parked at the tip of 'pu' after merging
all the other topics, as the test t9101 seems to expose breakage
in "git pull . remotes/git-svn".