2013/11/2 Andy Bradford <[email protected]>: > I believe steps 1--4 are fine and do work as expected. Due to the way > the http sync code is written, I don't know how many times we can > actually continue to prompt after a failure. At the moment, if > the password fails---whether stored or provided---there is only one > additional attempt allowed before the entire operation fails. (I don't > actually see a problem with this).
Agreed. A single retry is enough for me too. >> 5) If the clone/sync/... succeeded and a password was used which >> is different from the saved password, and no --once option is >> given, prompt whether the password should be saved. > > This would seem to be new behavior that does not, as of yet, exist in > Fossil. At the moment, the password is saved regardless of the > success/fail status of the sync operation (except when --once is in > force). Ideally, the password should be saved only when the authentication succeeds, success of the complete sync is not necessary. But I'm satisfied with the current behavior of saving it always. > This looks fine and is probably more correct behavior in the long term. > If we want to go this route now, we can morph the changes that I've been > making into changes for this newer behavior; in which case I think I > would like to move the changes out of pending-review and into a new > branch to work on them there. > > Otherwise, we can merge the current changes---which are complete as far > as I can tell with respect to the originally reported problem---and > begin a new branch to address the problem of saving the password only > *after* the sync has succeeded. I just made a little modification preventing the prompt if the saved password is already equal to the to-be-saved password. I will do some more testing the coming days, but I think the "pending-review" branch (another name would be better....) works fine as I expect. 2013/11/2 Andy Bradford <[email protected]>: > By the way, clone does not currently have a --once: > > $ fossil clone --once http://tester:testing@localhost:8081/ clone.fossil > unknown repository: --once For consistency it maybe should have one, but a --once clone should probably set auto-sync mode off: without a saved url, auto-sync doesn't make much sense. But I agree it's a separate issue. Let's not give "fossil clone" a --once option, I don't think I would ever use it. Andy, many thanks for all your work! I will do some more testing, but I think it really is an improvement, reducing the "surprise" rate of fossil. ;-) Regards, Jan Nijtmans _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
