response

doc/bugs/unable_to_access_annexed_files_from_a_git_repo_website_serving_via___34__smart_HTTP__34__/comment_1_38e521eae22d333bd4d68dfcca6c4339._comment

@@ -0,0 +1,24 @@
+[[!comment format=mdwn
+ username="joey"
+ subject="""comment 1"""
+ date="2018-10-04T17:52:43Z"
+ content="""
+So your web server is behaving differently when the user-agent contains "git"
+than when it does not.
+
+I don't think that git-annex can possibly talk to the git-http-backend
+successfully, since that is only going to support whatever subset of the
+http protocol that git smart http needs, and probably does not serve up the
+annexed files at all. When I force the user agent to not contain "git", it works.
+
+(It does look like git-annex init did not pass a user-agent when getting
+the config file, but with git-annex head I've verified it does pass it.)
+
+That leaves only the difference between http://datasets-dev.datalad.org/labs/haxby/raiders/
+and http://datasets-dev.datalad.org/labs/haxby/raiders/.git/  
+Now, with dumb http the first url would not work; git does not probe for
+$url/.git or $url.git when cloning over dumb http.
+
+I'm not clear if somehing in your rewriting made that url work, or if
+git-http-backend finds the repository.
+"""]]