On Thursday 24 August 2006 07:38, NextGen$ wrote: > * Ed Tomlinson <edt at aei.ca> [2006-08-24 07:25:16]: > > > On Wednesday 23 August 2006 15:19, Matthew Toseland wrote: > > > > > - Don't start the updater if the wrapper is broken > > > > I have problems with this one. I do not run the wrapper - I do want > > freenet to download new stable versions and then quit. > > I have the java command that starts freenet in a loop and this suffices to > > update freenet with much of the wrapper's complexity. > > It worked fine. > > > > I would not object to an 'are you sure' message but to block the operation > > is not reasonable. Without the wrapper its reasonable > > for someone to want freenet to download new jars and prep them for > > execution when the user restarts. > > > > Why do we _need_ this reduced flexibility? > > > > Ed > > We can't detect whether the user is running his node in a shell loop > or not : If he isn't the node would have just stopped and wouldn't > have "spread" the new update, hence I've added a check to block the > updater if we haven't detected the wrapper. > > Maybe we could have some sort of override... I'm not sure it worths > it though, we recommend using the wrapper, our installer installs the > wrapper... if you choose not to use it you're on your own.
I have had lots of 'fun' with the wrapper on amd64, which is why I try to avoid it. As it stands now, if you are not using the wrapper and from the web ask freenet to update, it says its updating and it could take 10 minutes. Watch on the console I see the message that it failed... IMHO what it should do from the web is tell me it cannot automate the restart and ask what it should do. From the console (or auto update) it should just update and shutdown. It would be fine to output message during startup or after a parm change saying it freenet cannot autorestart.