Hello, It isn't possible to do fossil sync/pull/push --once anymore after cloning a repository if the Fossil user does not match the local user due to this change:
http://www.fossil-scm.org/index.html/info/64aa75260f48781c This is what happens... Here, the new URL is properly parsed and stored in g.url: http://www.fossil-scm.org/index.html/artifact/912d6a207a3e070b?ln=123 Then a few lines later it calls user_select(): http://www.fossil-scm.org/index.html/artifact/912d6a207a3e070b?ln=130 Because the user that is now setup in the database does not match the local user, and no other checks in user_select() match, it falls through to attempting to load the URL calling url_parse(0,0) http://www.fossil-scm.org/index.html/artifact/58a793626f84fb38?ln=379 which gets parsed and then overwrites g.url with the stored URL. This results in a failure to sync to the desired URL. Notice here that it tried to sync to 8080 when it should have tried 8081: $ fossil sync --once http://remote:8081/ cannot connect to host remote:8080 Sync finished with 0 bytes sent, 0 bytes received This is because it wasn't able to identify the user---the user in the database does not match my local login, so user_select() tries to parse the URL stored in the database to use that instead). Prior to [64aa75260f48781c] it was possible to sync to any remote URL. I actually would like to revert [64aa75260f48781c] but there are others that prefer this behavior. :-) Thoughts for a fix? It seems onerous to require people to add an additional user that matches their local login to be able to sync to other instances of the same project. Perhaps when cloning both a default user that matches the URL user *and* a user that matches the local login should be created? Thanks, Andy -- TAI64 timestamp: 400000005345ae91 _______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
